DOT/FAA/CT-94/108 Review of the LiteratureFAA Technical Center Related to Screening AirlineAtlantic City International Airport,N.J. 08405 Passenger Baggage

OTICtSELECTEDEC b3 1994

FD

:,:•::,:;October 1994

7Ths dwunaen has been (appxomwedInterim Report f 0 or ub•li 1e•-e ýad ia!e; itt

distriibutio= is 7uniirnitec.

This document is available to the publicthrough the National Technical InformnationService, Springfield~, Virginia 22161.

U.S. Department of TransportationFederal Aviation Administration

CTS

NOTICE

This document is disseminated under the sponsorship of the U.S.Department of Transportation in the interest of information exchange. TheUnited States Govemment assumes no liability for the contents or usethereof. The United States Government does not endorse products ormanufacturers. Trade or manufacturer's names appear herein solelybecause they are considered essential to the object of this report.

Technical Report Documentation Page1. Rep No. 2. Govemrnment Accession No. 3. Recipient's Catalog No.

DOT/FAA/CT-94/1084. Title and Subtitle 5. Report Date

REVIEW OF THE LITERATURE RELATED TO SCREENING AIRLINE October 1994PASSENGER BAGGAGE

6. Performing Organization CodeACA-400

7. Author(s) 8. Performing Organizaton Report No.

Ronald J. Lofaro, Ph.D., Gerald D. Gibb, Ph.D.,Daniel J. Garland, Ph.D., Mr. Patrick C. Guide,Mr. James S. Baker, Mr. Mark St. Laurent

9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)

Federal Aviation AdministrationTechnical CenterAtlantic City International Airport, NJ 08405

11. Contract orGrant No.

FAA Grant 93-G-03712. Sponsoring Agency Name and Address 13. Type of Report and Period Covered

U.S. Department of Transportation FinalFederal Aviation AdministrationAssistant Administrator for Civil Aviation SecurityWashington, DC 20590

14. Sponsoring Agency Code

ACS-115. Supplementary Notes

Prepared by: Center for Aviation/Aerospace ResearchEmbry-Riddle Aeronautical UniversityDaytona Beach, FL 32114-3900

16. Abstract

The purpose of this research is to identify the underlying abilities and traits requiredfor successful performance as an airline passenger security screener. The primary focusof this effort is to develop a validated selection battery and protocols that can beeasily administered across a wide dispersion of organizations in the aviation securityindustry. Aside from the obvious costs associated with recruiting, selecting, andtraining replacement employees, there is likely to be a detrimental impact on theeffectiveness of airline passenger screening when a substantial percentage of theworkforce are novice workers. This research program seeks to identify the causes ofemployee satisfaction/dissatisfaction and develop appropriate interventions to curbexisting retention problems. A final objective of this program is to establish criteriathat discriminates levels of effectiveness among screeners. An effective and reliableperformance measurement system is required for this effort, and can provide usefulcriteria for other programs responsible for the establishment and validation of screenertraining, evaluation of advanced hardware, and implementation of screener performanceevaluation systems.

17. Key Words 18. Distribution Statement

Airline baggage screeners, personnel, This document is available to the publicperformance, selection, airline passenger through the National Technical Informationsecurity screener/pre employment inventory, Service (NTIS), Springfield, Virginia,APSS/PI, training, turnover, vigilance, 22161.performance measures, human factors

19. Security Classif. (of this repon) 20. Security Classif. (of this page) 21. No. of Pages 22. Price

Unclassified Unclassified 46

Form DOT F1700.7 (8-72) Reproduction of completed page authorized

TABLE OF CONTENTS

Accesion For

NTIS CRA&M PageDTIC TAB UUnannounced 0

EXECUTIVE SUMMARY Justification v

By-------.......1. BACKGROUND AND SIGNIFICANCE Distribution I 1

Availability Codes1.1 Purpose of this Research Avail aridior 1

PuroseDist Special

2. INTRODUCTION 2

3. REVIEW OF THE RELATED LITERATURE 4

3.1 Passenger and Carry-On Item Screening 43.2 Airline Passenger Screening Personnel Performance Issues 53.3 Selection and Hiring Procedures 8

3.3.1 Current Practices in Passenger Screener Personnel Selection 93.3.2 Previous Research on the Selection and Hiring of Airline Passenger

Security Screeners 10

3.4 Airline Passenger Security Screener Training 17

3.4.1 Syllabus Topics 173.4.2 Research on Training to Improve Screener Performance 19

3.5 Screener Turnover Rates 21

3.5.1 Incentives/Reward Systems 22

3.6 Vigilance Operational Significance for Airline Passenger Security Screeners 22

3.6.1 Origins of Vigilance Research 233.6.2 Theories of Vigilance 233.6.3 Vigilance Decrement 253.6.4 Vigilance Tasks 263.6.5 Performance Measures in Vigilance-Related Tasks 273.6.6 Techniques to Combat the Vigilance Decrement 28

iii

3.7 Selection Test Development for U.S. Army Personnel 29

3.8 Effects of Stressors on Sonar Operators 30

4. HUMAN FACTORS ISSUES IN AIRLINE PASSENGER SECURITYSCREENING CHECKPOINT WORKSTATION DESIGN 31

4.1 Lighting and Glare 324.2 Floor Grade 324.3 Workstation Architecture 324.4 Communication and Data Transfer Systems 334.5 Further Considerations 34

5. OTHER SOURCES 35

LIST OF ILLUSTRATIONS

Figure Page

1 The Effectiveness of Passenger Screening 6

LIST OF TABLES

Table Page

1 Comparison of Screeners Nationwide to Veteran Screeners 16

2 ATA-Approved Classroom Syllabus for Checkpoint Security Screeners 18

3 Starting Wages and Benefits for Airline Passenger Security Screeners by Airport andSecurity Contractor 22

4 Partial Listing of Factors that Affect Performance in Vigilance Tasks 27

5 Workstation Evaluation at San Francisco International Airport (SFO) 34

iv

EXECUTIVE SUMMARY

"4The purpose of this research is to identify the underlying abilities and traits required forsuccessful performance as an airline passenger security screener. The primary focus of this effortis to develop a validated selection battery and protocols that can be easily administered across awide dispersion of organizations in the aviation security industry. We expect a wide range ofproducts to emerge from these efforts that can include performance tests, screener interviewertraining, and interviewing protocols.

First year efforts however also indicated that a severe employee turnover problem exists. As aresult, initiatives to identify factors contributing to job satisfaction and career retention wereincorporated. Understanding the causes of employee turnover and developing viable solutionsbecame an integral component of this project.

A series of three reports contain our first years' efforts and focuses on the literature review,acquisition and development of subject matter expert (SME) data, and job task analyses (JTA).The first report presents the background and literature review. It contains a summary of priorwork related to the selection of screeners, selection work in related occupations, application ofconcepts in vigilance research, human factors problems, and issues in checkpoint design. Thesecond report describes the results of the job task analysis and concepts related to thedevelopment of a prototype selection test battery. The final report in the series examines theissues related to employee turnover. This report discusses the Delphi analysis used to identifyjob satisfaction and dissatisfaction factors.

SME data were generated through interviews during the JTAs and modified Delphi techniquesusing 115 screeners. Concurrent with the development of the SME database, JTAs wereconducted at various security sites across the nation. Considerable attention was given towardidentifying a representative sample of each site. Integration of these results will lead to thedevelopment of a taxonomy of abilities and traits that have potential as predictors of screenerperformance. In addition, causes of self-initiated employment termination will identify issuesand factors that have to be addressed to retain effective airline screening personnel.

Second year efforts will primarily focus on the initial validation of a prototype selection testbattery and the development of intervention techniques to improve career longevity. It isexpected that a considerable portion of this phase will be devoted to validating constituent testswith strong psychometric characteristics and establishing reliable performance criteria using falseimage projection technology. Data relating to critical psychometric properties (i.e., differentialstability, task reliability) will be collected within a university laboratory environment.Concurrently, Klein and Associates, in conjunction with our efforts, will develop and validate aperformance measurement system that discriminates effective screeners from those whodemonstrate performance problems. False image projection during on-line operations has beenchosen as the most pragmatic and useful performance criteria to assess the predictor variables inthe validation efforts. This methodology offers the best approach to distinguishing levels ofperformance among screeners. Field testing efforts during the second year will concentrate on

v

initial validity studies, using new hires shortly after the company application process has beencompleted. New hires will be administered the prototype tests during the application and pre-employment process, but before training is initiated.

Several intervention programs to mitigate turnover rates will be developed during this phase ofthe program. As intervention programs are implemented at various facilities, average employeetenure rates, employee turnover, passenger/aircrew complaints and other criteria will be trackedto assess the impact of the interventions. In conjunction with this effort, we will be conducting anationwide survey to scale the items that resulted from the Delphi analyses. Scaling techniqueswere selected to develop relative weights for each of the satisfiers and dissatisfiers that emergedduring the Delphi workshops. In addition, the data will assist in assessing the generalizability ofthe findings from the Delphi analyses.

Additional data will be collected at select sites using the Likert Characteristics of OrganizationalClimate survey to assess the divergence between how screeners perceive their current workenvironment and how they would like it to be. These data will also be used in developingintervention programs designed to reduce the high turnover rates.

vi

1. BACKGROUND AND SIGNIFICANCE.

The safeguarding of civil aviation against hijackings and/or sabotage has been a grave concern togovernments around the world. Such malicious acts have had profound detrimental effects oncivil aviation (e.g., the bombing of Pan Am flight 103 over Lockerbie, Scotland). The cost ofsuch acts, when considering the loss of lives and disruption to air carrier operations, isunacceptable.

Since the inception of the passenger and carry-on baggage screening program in 1973, over 11billion passengers and their carry-on items have passed through airport security checkpoints.According to the Federal Aviation Administration (FAA, 1991), this has resulted in the detectionof more than 45,600 firearms and over 20,150 arrests.

Periodic inspections evaluating the effectiveness of airline passenger security screening arereported to average about 90 percent for success in detecting "test weapons" submitted by FAAinspectors posing as passengers. However, a 1987 General Accounting Office report (GAO,1987a) found that the nationwide success rate in such tests was only 80 percent. Moreimportantly, their analysis found tremendous variability between security facilities in detecting"test weapons." One site had a "hit" rate (i.e., success rate) of only 34 percent. The GAO reportclearly demonstrates the lack of standardization and effectiveness among many security facilities.

In a report by the President's Commission on Aviation Security and Terrorism (President'sCommission, 1990), the commission members noted the importance of adequate FAA guidancein the selection of security screeners. The commission was critical toward the FAA with regardto how little attention was paid on recruiting and motivating security personnel.

To date there has been little emphasis placed on the selection of security screeners. The AirTransport Association (ATA), in recognizing this shortfall, provided input and limited fundingfor the development of a 32-item questionnaire to assess specific attributes (e.g., attentiveness,dependability, attitude). The breadth of this instrument is quite limited however, and the validityis unknown. An analysis of this instrument is provided in a subsequent section of this report.

These concerns lay the foundation for this research. Standardized screening and selection ofsecurity personnel provide a means for hiring a high quality workforce while reducing attrition ofthose suited for the profession. In addition, selection of potential personnel can be accomplishedusing research-based algorithms wherein the probability of a successful hire can be estimated. Auniform approach to selecting security personnel, based on aptitude and ability, further offers thepotential to developing standardized training programs. This would be possible since trainingobjectives and content could be directed at a more homogenous candidate population with therequisite baseline abilities and skills.

1.1 PURPOSE OF THIS RESEARCH.

In recognizing the critical need for an effective and stable airline passenger security screeningworkforce, this research is directed toward developing a screening program that identifies and

1

staffs the nation's airline security system with the most capable and motivated individuals. Thesuccess and deterrent potential of that system is primarily dependent on the personnel whooperate it. An effective selection process for security personnel that is standardized nationwidewill serve to enhance the overall capability and deterrent potential of the entire airline securitysystem.

A priority that emerged during the first year's work was the critical need to address the problemof extraordinarily high employee turnover rates and training attrition rates. Aside from theobvious costs associated with recruiting, selecting, and training replacement employees, there islikely to be a detrimental impact on the effectiveness of airline passenger screening when asubstantial percentage of the workforce are novice workers. This research program seeks toidentify the causes of employee satisfaction/dissatisfaction and develop appropriate interventionsto curb existing retention problems.

A final objective of this program is to establish performance criteria that discriminates levels ofeffectiveness among screeners. An effective and reliable performance measurement system isrequired for this effort, and can provide useful criteria for other programs responsible for theestablishment and validation of screener training, evaluation of advanced hardware, andimplementation of screener performance evaluation systems.

2. INTRODUCTION.

A successful national airline passenger security screening program is predicated on effectiveFAA guidance in the selection of screeners and in their training, as well as on the importancewhich individual airlines place on security. The purpose of passenger and carry-on itemscreening is to ensure that crimes against civil aviation are minimized and ultimately eliminated.In order to achieve this, the security team is provided with relatively sophisticated equipment

that requires a thorough understanding of its capabilities; but more importantly, they must be ableto operate the equipment effectively under varying workload conditions (e.g., peak periods orairport "rush hours"). As with most safety-critical systems, there is no room for system-inducedor operator-induced errors.

Criminal acts against civil aviation have led to a concerted effort to strengthen the internationalaviation security system. In response to a series of crimes against civil aviation, the U.S.Congress mandated, through amendments to the Federal Aviation Act of 1958, that the FAAassume primary responsibility for civil aviation security. In accordance with the act, the FAA, incooperation with airports and air carriers, is responsible for ensuring that security programs andprocedures are instituted that will safeguard passengers, crew, aircraft, and airports. As part ofits mission, the FAA Office of Civil Aviation Security establishes security requirements, inspectsairline and airport security operations, and issues civil penalties for noncompliance with therequirements.

The Air Transportation Act of 1974 requires that all passengers and property intended to becarried in an aircraft employed in interstate transportation be screened by weapon-detectionprocedures or agents of the air carrier. Section 315 of the Federal Aviation Act directs the FAA

2

to prescribe regulations requiring the screening of all passengers and carry-on items for thepresence of unauthorized items. Section 316 of the Act also requires regulations to protectpersons and property aboard aircraft from acts of criminal violence and piracy. Federal AviationRegulations (FARs) Parts 107 and 108 govern domestic airport and air carrier security,respectively. These regulations mandate the adoption and effective implementation of minimumsecurity programs by airports and air carriers. These programs must include specific measuresfor passenger screening, protection of aircraft, and airport access controls. In general, FAR Part107 requires that the airport operator: (a) create a security program for the airport, (b) providecontrols to prevent or deter unauthorized persons from accessing the air operations area, and (c)provide law enforcement support. For air carriers, FAR Part 108 generally requires that thecarrier: (a) adopt and carry out a security program, (b) screen passengers and property, (c)provide and use ground and in-flight security coordinators, and (d) prohibit unauthorized accessto the airplane.

Indirect cargo air carriers (ICAC) are required under FAR Part 109 to adopt and carry out asecurity program approved by the FAA. This security program is intended to prevent and deterthe unauthorized introduction of explosives and incendiaries into packaged cargo and mailtendered in air commerce. The U.S. Postal Services (USPS) and its military counterpart, theMilitary Postal Service Agency (MPSA), maintain significant jurisdictional and practical controlover domestic and international airmail carried by U.S. air carriers. The FAA is currentlyworking with the USPS and MPSA to review airmail security and screening procedures.

Air carriers are responsible for screening passengers with metal detectors and X-raying theircarry-on items for weapons and explosives. Air carriers have generally elected to contract withprivate security firms to perform security functions. At the time of this writing, there were over15,000 screeners employed by some 46 security firms nationwide.

The air carriers will frequently hire the security firm with the lowest bid (St. John, 1991).Nonetheless, the air carrier is held responsible by the FAA for the effectiveness of the screeningprocedures, and both the air carriers and the FAA monitor a security firm's performance. Aircarriers are responsible for developing security plans which assure prevention or deterrence ofaircraft hijacking, sabotage, and related criminal acts. The airport operators are to maintainsecurity programs which prevent and deter unauthorized entry into air operations areas andprovide law enforcement support for the screening system and overall airport securityrequirements (FAA, 1991).

The FAA's role in aviation security expanded significantly in 1985 with passage of Public Law99-83, the International Security and Development Cooperation Act, with further increases inFAA involvement in 1990 with the passage of PL 101-604 (the Aviation Security ImprovementAct), as a result of the downing of Pan American flight 103 over Lockerbie, Scotland. The Actrequired the FAA to assess the adequacy of security at foreign airports served by U.S. carriers, andthe security procedures of foreign air carriers flying to the U.S.

3

3. REVIEW OF THE RELATED LITERATURE.

.3.1 PASSENGER AND CARRY-ON ITEM SCREENING.

The most visible aspect of domestic airline security is the screening of passengers and carry-onitems. For all practical purposes, the focus of the security procedures for domestic flights is todeter hijackings and has been so since their inception in 1973. The GAO (1987a) report foundthat for a given six month period (July 1, 1989 to December 31, 1989), over 535 million personswere processed through screening checkpoints at U.S. airports. This process resulted in thedetection of 1,464 firearms, of which 1,406 (96 percent) were detected by X-ray inspection, 30 (2percent) by physical search, and 28 (2 percent) by use of metal detectors. In addition, 10explosive/incendiary devices were discovered during the period including five grenades, threefireworks, one flare gun, and one tear gas device. During this six month period, 764 personswere arrested at passenger screening points for unauthorized carriage of firearms orexplosives/incendiary devices.

The FAA's testing of the effectiveness of the screening process is straightforward but somewhatsimplistic. The agency has historically used test items, such as simulated dynamite tied togetherwith a large clock and attached wires, to test the X-ray equipment and the ability of the operatorto detect a potentially lethal weapon. The test item is generally placed in a brief case or bag withlittle effort to conceal or disguise it. The briefcase or bag is then taken by an FAA securityinspector posing as a passenger, to the passenger screening point and submitted to the screeningprocess.

One would expect near 100 percent success rate in detecting "test weapons" since the FAA testobjects resemble only very obvious threats and since the FAA inspectors are not allowed to hidethe test object(s) into the carry-on (GAO, 1987b). Furthermore, screeners are often aware thatthey are being tested because FAA inspectors are well known to screeners at most airports. Oneinvestigator (Vosburgh, 1993) contends that this approach is not very practical. He noted thatmost screeners "... would probably not recognize a real bomb [or threat] since they trainprimarily to identify test objects similar to those used during airline and FAA tests" (p. 61).

The FAA reports that, nationwide, the screening process is identifying FAA test weapons slightlyover 90 percent of the time (FAA, 1991). The screening system's performance has improvedsince 1987, when the GAO noted that the tests found an average detection rate of about 80percent (GAO, 1987a). These percentages are based on 2,419 screening tests conducted at 28major airports (i.e., those that screen over 2 million persons annually) from September, 1986 toDecember, 1986 (GAO, 1987a).

Airline passenger security screeners rely on relatively sophisticated equipment consisting ofwalk-through metal detectors and X-ray inspection systems to screen passengers and their carry-on items. Hand-held metal detection devices are also used, but primarily as backup support forwalk-through detectors. In addition, screeners may require physical searches for items insuspicious carry-on baggage. Each component of the process-X-ray, metal detector, andphysical search-is periodically tested by the airline and FAA. While there have been some

4

technological improvements to screening equipment, the process generally operates the sametoday as it did when implemented in 1973. Nonetheless, some industry critics (St. John, 1991)report that the airlines put their trust in "... metal detectors and X-ray machines-spendingmillions on equipment and little on the people who operate it" (p. 85). Regardless of the amountspent and the justification for such advanced equipment, Hughes (1993) noted that a keyconclusion of the latest National Research Council study is that there is no single detectiontechnology available today that can provide a high probability of bomb or weapon detection.

There is an effort however, in the FAA's Research and Development Program, to enhance andautomate X-ray systems used in the screening of passengers and carry-on items. Severalmanufacturers are independently showing significant innovation in extending X-ray technologyto identify specific threats. Work is underway to improve concourse X-ray system performanceby concentrating on the development of automatic pattern recognition software and hardware.Integrated into current X-ray detectors, automatic pattern recognition systems could alert theoperator to suspicious items in luggage by graphically highlighting the suspect item.

3.2 AIRLINE PASSENGER SCREENING PERSONNEL PERFORMANCE ISSUES.

On December 21, 1988, an explosive device was detonated aboard Pan American (Pan Am)Flight 103 en route from London to New York resulting in the deaths of all persons on board and11 persons on the ground in Lockerbie, Scotland. As a result of this tragedy, the President'sCommission on Aviation Security and Terrorism was formed on August 4, 1989, by ExecutiveOrder 12686 (President's Commission, 1990). The Commission's objective was to conduct acomprehensive study and appraisal of the practices and policy options to prevent terrorist actsagainst civil aviation.

In a recent study, Vosburgh (1993) noted that a combination of errors on the part of the carrierand their security contractor contributed to the destruction of Pan Am Flight 103. Vosburghcited some of the contributing factors; they included: "... passenger screening; baggage handling,especially matching passengers and their bags; [and] almost complete reliance on X-rayequipment to find explosive devices ... " (p. 5).

The importance of having a consistent set of selection and training standards for airport securitywas demonstrated by the Pan Am 103 investigation. The investigation suggested that thesecurity deficiencies found could be connected to breakdowns in airline security performance. Inaddition, the investigation found that Pan Am security personnel failed to properly screen 38passengers at Heathrow airport (President's Commission, 1990).

During the late 1960s and early 1970s, U.S. air carriers had experienced a sudden and unexpectedincrease in domestic aircraft hijackings (see figure 1). The FAA responded with a coordinatedand extensive civil aviation security system whose primary deterrent was passenger screeningusing metal detectors and X-ray machines. Figure 1 illustrates the effectiveness of these securityprocedures by comparing the hijacking statistics before and after the onset of passenger screeningimplemented in January of 1973.

5

Another study briefly discussed in "Study Tallies" (1991) examined hostile acts against civilaviation between January 1980 and December 1990. During that 1 1-year period, theinvestigators uncovered 304 hijacking incidents-one every 13 days. Foreign air carriershowever, encountered the majority of these hijackings.

40 40

35 - 35

. 30 30

S25 25

" 20 - 20

S15 15

.. 10 10

5 5

0 0

1969 1970 1971 1972 1973 1974 1975 1976

Calendar Year

FIGURE 1. THE EFFECTIVENESS OF PASSENGER SCREENING. (ADAPTED FROM"SCREENER TRAINING PROGRAM INSTRUCTOR'S GUIDE" BY AIR TRANSPORT

ASSOCIATION OF AMERICA (ATA) IN COLLABORATION WITH THE FAA.)

Aeroflot had suffered the most with 24, followed by LOT Airlines with 16, and Iran Air with15. Although hijacking incidents do not appear to be an immediate threat for U.S. operators,St. John (1991) stated, "As international security tightens, the United States, with its lack ofsafeguards at airports, becomes more inviting to terrorists" (p. 85).

Oster, Strong, and Zorn (1992) reported that the risk of a terrorist incident or sabotage act isalways present and "... is not confined to the middle east and other trouble spots in the world"(p. 143). Supporting this position, Oster et al. investigated the number of explosions aboardaircraft by geographic region between 1950 and 1989. They found a total of 72 incidents, withAsia experiencing 20 (28 percent), North America with 18 (25 percent), Western Europe with15 (21 percent), the Middle East with 9 (13 percent), and the remaining 10 in South America,Africa, and Eastern Europe.

6

The FAA has initiated several programs to counter terrorist threats. The FAA sponsors aviationsecurity training programs at the Transportation Safety Institute (TSI) in Oklahoma City,Oklahoma under the authority of Section 316(c) of the Federal Aviation Act of 1958. Thecurrent programs relating to aviation safety and security are administered by the AviationSecurity Division (ASD) of TSI. The FAA identifies in its security program the corerequirements and guidance for the initial, recurrent, and on-the-job training of airline passengersecurity screeners at domestic airports. However, several task forces and studies have foundthat the quality of this training varies widely among the airlines.

The ATA has recognized the need for improvements in the selection and training of screeningpersonnel. The ATA encourages air carriers to conduct tests for screeners on a regular basis.However, these tests use the identical testing objects used by the FAA inspectors. In 1989, therewere 56,000 tests performed by the air carriers with a reported 96 percent detection rate (ATA,1990). However, the validity of these tests remains questionable. Often screeners recognize theair carrier personnel who conduct the tests and are alert to the likelihood that a test will beconducted. Actual threats typically occur without warning and without salient cues present. Inorder to further improve on this performance, the ATA (1990) has developed a pre-employmentinventory for selecting security screeners. However, this test instrument-the Airline PassengerSecurity Screener Pre-Employment Inventory (APSS/PI)-has received limited implementationand its validity is also unknown.

The ATA has developed programs for hiring and retaining quality personnel at securitypositions. These include background checks, aptitude tests, psychological screening, Englishlanguage proficiency, and wage surveys. The FAA has since adopted many of these guidelines.In addition, personality profiles have been developed by the ATA for the security field, reportingthat nearly two-thirds of security personnel applicants can be disqualified because they do not fitthese presumed profiles (ATA, 1990).

In 1978, following unsatisfactory test results of the security screening process, a task force of theFAA and airline security personnel studied ways to improve performance at passenger screeningcheckpoints. This task force's report, referred to as the "Human Factors Study," recommendedseveral actions which were eventually endorsed by both the FAA and the airlines. For the mostpart, these recommendations focused on the personnel-related aspects of the process such as highemployee turnover rates, low pay, and inadequate training. Although the FAA and the airlineindustry endorsed the study's recommendations, the air carriers have been slow in fullyimplementing them.

In a 1987 GAO report to the Secretary of Transportation (GAO, 1987b), it was reported that aninvestigation of screening processes at six major airports found that many of the problems thatwere addressed in the 1978 human factors study, in fact, still existed. For example, security firmmanagers reported that screeners were still being paid at or near minimum wage and that low paycontributes to high-turnover rates-in some cases, about 100 percent annually-and furtherresulted in problems in hiring capable people.

7

The responsibility for establishing an effective and integrated aviation security program is sharedby the air carriers, airports, Federal, State, and local governments (FAA, 1991). In anunpublished study, Thaher (1991) reported that the FAA spent $10 million in 1990 on thedevelopment of new passenger screening, explosive detection, and baggage inspectiontechnology. However the aviation security industry, more specifically the airlines, have beenaccused of a penny-pinching attitude toward security (St. John, 1991). Furthermore, it has beensuggested that only 65 cents per passenger ticket goes toward security, in contrast to thepassenger facility charge (PFC) which can top out at $3.00 per passenger ticket. These limitedfunds obviously affect the quality of the equipment obtainable as well as the quality of personnelselection and training. "The last thing airlines want to consider is spending more money forsecurity. What they do not seem to realize is that security must be a component part of theirservice, just like on-time flights, safety, and passenger satisfaction" (Vosburgh, 1993; p. 6).

The International Civil Aviation Organization (ICAO) has also identified a number of areas inaviation security requiring attention (Coninsx, 1993). These include: low wages, lack of careerprospects, lack of challenge in the job, lack of authority, poor working hours, pressure from theairlines to perform during peak hours, fear of possible health effects (e.g., bomb detonation,radiation exposure), penalties for poor performance, and lack of sufficient knowledge to handledangerous situations (e.g., armed passengers, explosive devices). We must first create betterworking conditions and better salaries in order to attract suitable personnel. When appropriate,we must also reward employees for their efforts, loyalty, and achievements.

It is impossible to overstate the importance of selecting high-caliber security personnel. Wageconstraints cannot be accepted as a valid reason for hiring a substandard security force. Adequateemployee screening is a vital prerequisite to an effective long-term security screening operation.After employees are hired, their performance should be realistically critiqued with regularperformance evaluations as done with other aviation occupations (e.g., pilots, air trafficcontrollers).

3.3 SELECTION AND HIRING PROCEDURES.

The selection process of those persons who have the necessary attributes for target detection isrelatively unexplored. Selection procedures typically assess candidates in terms of test scoresthat serve as predictors of job success (Atkinson, 1973). The purpose of personnel selection is toidentify the most suitable applicants, train them adequately, and ensure they retain theiremployment within the organization. Who are the individuals that are best able to detectpotential threat items? Who are the individuals that are better able to maintain attention over along duration of time? Are there similar personality characteristics of the good performers? Ifso, what are they? Can airline passenger security screeners be identified as successful performersthrough a series of tests, attributes, and/or biographical information?

Coninsx (1993) noted that a precise personnel specification, or candidate profile, is required foran efficient recruitment process; however none exist for airline passenger security screeners. Inorder to be able to carry out the job to the required performance standards, the personnelspecification identifies the most important areas of knowledge, skills, and personality

8

characteristics required. Drawing up these specifications is not easy. They often includerequirements on physical abilities, general intelligence, interests, disposition, and motivation(Coninsx, 1993). In addition, the requirements need to be adapted to the evolution of thetechnology utilized. For example, recent advancements in X-ray displays utilize color to detectvarious hazards, thus those screeners who can not discriminate specific colors should be screenedout of the pre-employment selection process. This, however, raises some legitimate concerns forthose screeners who have been on the job for years prior to implementing these new displays.

Some investigators (Pierce, Crumley, and Clifford, 1991) report that few guidelines exist forselecting personnel who are best suited to perform vigilance tasks (e.g., X-ray screening). Pierceet al. suggest that there is potential for improving target detection performance by creating a scalewhich can accurately identify those candidates who are vigilance maintainers. In addition, otherinvestigators (Boff and Lincoln, 1988) emphasize the need to develop tests that could correlatevigilance performance with membership in a class or group.

Even the most advanced security equipment employing the latest technology can become uselessin the case of human error, human negligence, and personnel inefficiency. Some investigators(Coninsx, 1993; Drury and Fox, 1975; Funke, 1979; "Keeping the operator," 1992; Vosburgh,1993) purport that the human has proven to be a weak link in inspection systems. Nonetheless,detecting targets in low signal-to-noise environments where search time is limited can, at times,be a difficult task, particularly when the signals (i.e., targets) are intermittent, unpredictable, andinfrequent.

Screeners must always maintain a high level of alertness so that they do not miss any potentialthreats. Sustaining high levels of attention is critical for such related occupations as air trafficcontrol, nuclear power plant monitoring, industrial inspection, and piloting an aircraft (Edwards,1990). For example, detecting the components of a bomb (i.e., timing device, initiator, powersource, explosive) as seen through an X-ray requires attentive behavior. The screener must then,during the brief presentation of the image, identify the components and determine if they are partof an obvious threat. The components must be identified for what they are, and then associatedwith the explosive device. Since the parts, whether explosive or weapon elements, can be spreadout over several bags, the process can be quite difficult (Vosburgh, 1993).

3.3.1 Current Practices in Passenger Screener Personnel Selection.

The position of airline passenger security screener is relatively new, having been established asrecently as 1973. Unlike other aviation personnel (e.g., air traffic controllers, flight deck crew,and maintenance technicians), comparatively little attention has been devoted to developingstandardized selection protocols for this occupation. This can be attributed to a number offactors including: (a) relatively low salary scales; (b) lack of lengthy and costly training required;(c) minimal education and technical requirements for entry to the field; (d) no similar militarycounterpart; (e) lack of job complexity; (f) traditionally high turn-over rates; and (g) thedecentralization of hiring resulting from a host of private concerns employing these individuals.Private organizations do not have the resources, or the expertise to develop specialized selection

9

programs. Consequently, there is a marked paucity of personnel selection research that exists forthis particular occupation.

Indeed the only effort aimed at the development of an airline passenger security screenerselection system in the United States is a series of studies commissioned by the Air TransportAssociation (ATA). In 1988, the ATA contracted with an independent consultant group(hereafter referred to as the contractor) to develop selection protocols that could be widely usedby the ATA member organizations.

3.3.2 Previous Research on the Selection and Hiring of Airline Passenger Security Screeners.

3.3.2.1 Critique of the APSS/PI.

The contractor initially used a concurrent validation technique to construct their selectioninstrument, referred to as the Airline Passenger Security Screener/Pre-Employment Inventory(APSS/PI). The instrument focuses primarily on personality traits believed to be related to jobperformance. These traits were derived from a review of available job description informationand interviews with security screening experts. The basis of the contractor's wolk relied onadministering the Criterion Inventory Series (CIS) to over 200 screeners working in fivegeographically dispersed airports across the country. Those particular airports will not bediscussed in this report. Performance on the predictor instrument was compared with severaloperational job performance measures. These measures included: (a) FAA testing errors, (b)airline testing errors, (c) security firm testing errors, (d) supervisor ratings, and (e) unexcusedabsences or incidences of reporting late to the job site. The measures were chosen by ATASecurity Committee members and managerial personnel at the security firms for their availability(i.e., screener personnel files), their operational significance, and ease of measurement.

The contractor found a relationship between several of the predictor items and both objective jobperformance measures (testing errors) and personal reliability measures (absenteeism/lateness).No relationship was found between supervisor evaluations and the predictors or the objective jobperformance measures. The analysis of these data was used to construct the APSS/PI. TheAPSS/PI is then a modification of the CIS and includes within the items an "honesty" (reliability)scale. Elevated scores on the APSS/PI indicated a "risky" applicant and generated a "notrecommended for hire" classification. Those who scored high on the APSS/PI were found tomake significantly more screening errors, have higher absentee rates, and have poorer jobperformance ratings.

In 1990, the contractor conducted an evaluation study with an additional group of 211 screeners.Although several new test sites and participants were added to the study, it essentially remained aconcurrent validation similar to the 1988 effort. APSS/PI scores were compared with a numberof testing errors and job tardiness measures found in the personnel records of screenerparticipants for the preceding six month period. Months of employment and supervisory ratings(disciplinary problems, communication skills, attentiveness, dependability, and attitude) on a 5-point scale were also recorded. As previously found, a significant relationship was reported to

10

exist between APSS/PI scores and objective performance measures. Again, there was norelationship between APSS/PI and subjective evaluations provided by supervisory personnel.

There are a number of serious problems with this developmental effort. Of considerable concernis the use of concurrent validation to establish the predictive utility of the APSS/PI instrument. Inboth the 1988 and 1990 evaluations, APSS/PI "scores" were compared against historical data inthe employee's personnel records. The use of this approach is highly questionable, particularly inthe 1990 evaluation. Given the nature of the criteria and the use of periodic screeningperformance checks, why was there not a predictive validation conducted in 1990? Newscreeners could have been administered the APSS/PI, predictions generated, and theirperformance evaluated through the established techniques. Concurrent validation techniques donot assure predictive validation. The use of incumbent screeners, as opposed to applicants,further introduced the possibility of restriction-in-range in the sample. Data from the CIS for the1988 sample indicated only relatively moderate variability among individuals, for example.

The investigators further falsely assumed that the incumbent screeners had all the requisite traitsand abilities, to the exclusion of the possibility that others with different profiles might also besuccessful. The use of concurrent validation techniques also fails to account for most new hiresleaving the career field. This is an important concern given the typically high turn-over rates.The reliance on an existing, historical criterion with questionable psychometric characteristics isalso problematic. A more desirable approach should have included a standardized criterion thatcould have been developed and administered by displaying "targets" embedded in a video tapeconstructed for this purpose. This approach would have eliminated any possible contaminationfrom expectancy bias during the evaluations.

In these efforts concurrent validation was not a justifiable or suitable substitute for initial andcross-validation of the instrument. The studies are exploratory at best and do not constitute anacceptable validation as dictated in the 1978 employee selection guideline as the contractormaintains. A major issue regarding the development and validation of the APSS/PI centers onthe sample selection procedures. There is some indication that part of the sample used to selectthe aptitude items for the APSS/PI from the CIS was also part of the sample used to validate thefinal scale. The investigators reported that some of the same sites and security firms were used inboth studies, but no information is provided regarding if, and how many, subjects were part ofboth analyses. Constructing and validating a selection instrument with the same individuals isnot an appropriate method.

The development and content of the APSS/PI raise numerous issues as well. The instrument isapparently focused on personality traits alone and excludes the use of other possible domains ofpredictors (e.g., motivation, perceptual skills, cognitive abilities) that may contribute to screenerperformance. The limited scope of human characteristics that the APSS/PI measures wouldseverely curtail its ability to successfully predict performance. The most obvious omission was afailure to even consider the use of a work sample test. The contractor appears to have operatedunder the assumption that personality measures alone would have accounted for 100 percent ofthe variance in performance.

11

Exceedingly weak evidence is offered for using strong analytical methods to establish thepredictor variables or their construct validity. The study relied almost exclusively on a limitedsample of subject matter expert interviews and U.S. Department of Labor descriptions to derivepossible predictor variables. Some indication was given that the investigators simply identifiedpersonality characteristics that would lead to negative performance (i.e., errors, job tardiness)with no mention of any type of analysis. Additionally, no information or description of howthese data were collected, interpreted, and analyzed is presented. The basis for test developmentor selection of possible predictor variables was inadequate for this effort. Strong selectionprograms typically begin with a complete job or task analysis to define the underlying traits andabilities. The derivation of a taxonomy of human characteristics to perform as an airline screeneris almost devoid in these studies. This consequently led to the development of a narrowlydefined selection instrument.

The origination of the APSS/PI is itself problematic. While the validity and reliability of the CISis stated to exist, no data are provided to support this. The instrument is virtually unknown andalmost no description of its structure or development is offered. One is left wondering exactlywhat personality dimensions are assessed with the instrument.

The refinement of this instrument to the APSS/PI becomes even more puzzling. How were itemsselected from the CIS to establish the APSS/PI and why? Furthermore, there are no data to: (a)establish the reliability of APSS/PI; (b) demonstrate the construct validity of the scale; or (c)establish the accuracy, reliability, or validity of the embedded lie scales. No information wasever presented on how the "reliability" (lie) scale items were selected, or how these were evenused in the validation study. Were subjects eliminated from the validation study if they exhibitedelevated reliability (lie) scale scores, and what were the cut-off values? The performance datapresented reveals various degrees of freedom among the different variables indicating somesubjects were deleted from analysis. It is not known why were these subjects deleted.

The APSS/PI has been purported to measure the personality traits requisite of a successfulscreener. However, only a global score and a "lie" score are generated by the APSS/PIinstrument. No data are presented that indicate the multi-dimensional nature of the severalpersonality traits that were initially put forth as important. Nor was any consideration apparentlygiven to the possibility that individuals would exhibit differing degrees of personal traits, as thescale provides no value to each purported personality dimension. In short, no measurement orevaluation of individual traits was given. It is questionable if the APSS/PI is actually a compositemeasure of personality at all given the marked paucity of data available. The personalitydimensions that are purported to be embedded in the scale are not identified, described, ordelineated by the contractor investigators. Considering the limited nature of the APSS/PI (32-items), it is unlikely that sufficient items even exist as markers to adequately assess severalpersonality dimensions, particularly since only dichotomous responses (agree versus disagree)are possible.

Finally, given the shortfalls of the CIS and APSS/PI, why weren't established, well documented,and researched scales used to assess the personality traits of interest? The investigators provideno rationale for beginning this line of research with the CIS except for gratuitous arguments

12

against the lack of "customized honesty scales" that were not designed for the airport screenerpopulation. Yet no data or argument is provided to demonstrate that airport screeners aredifferent from the general population.

An additional area of concern is the use of factor analysis and the development of compositecriterion measures. Factor analysis is a data reduction procedure used to define the underlyingdimensions of a set of variables. It is not clear why the investigators chose to enter three sets oftesting errors, an absenteeism variable, and the five subjective supervisory variables into theanalysis. As expected, the testing error measures clustered together into one "factor" while thesupervisory ratings defined a separate distinct factor. Testing errors was a simple dichotomousmeasure of target detection versus non-detection. In this environment where the detection of"targets" is the key job performance measure embedded within identical task parameters, onlythe organization performing the administration of the performance check varied (i.e., the FAA,security company, airline).

There is no justification to assume the three measures were not related. Similarly, it could beexpected that the five supervisory ratings would also define a factor given the nature of the scaleused and the wealth of information in the literature regarding supervisor ratings and the haloeffect. There was no apparent need to identify an underlying factor, nor was one fully described.

The troublesome aspect of the factor analysis was the use of these results to construct compositemeasures. One composite measure of objective job performance summed the total of all FAA,security company, and airline testing errors into a total testing errors measure. However, the datato form this composite measure were historical records over a 6-month period using a simplescale that recorded the number of actual screening errors. One value on that scale includes acategory for six or more errors. Simple summation of error frequency counts across these scaleswould cause a restriction of range problem for the criterion by effectively building a ceilingeffect into one end of the error criterion. This is caused by equally weighting points on a scalewhich were not in fact equivalent. Of greater concern is the failure to use a mean measure oferrors. The base number of job performance checks is not provided leaving one to question ifthere were a different number of performance checks between screeners. Whether all theperformance checks themselves were comparable remains uncertain. Variability in the length ofthe checks, number of targets ratio, and target ambiguity would introduce considerable errorvariance.

The second composite objective performance measure, overall negative performance, adds thenumber of times late to work to the previous composite measure. The investigators offer nojustification/rationale for combining these two separate sets of variables. Furthermore, it appearsinappropriate to equate a screening error with arriving at work late. A simple summationprocedure weighs the two variables equally, however. It is plausible that a screener who is quiteproficient at target detection is consistently late for work. Yet it would be impossible todistinguish this individual from a screener who always appears on time for work but is less adeptat target detection, the actual operational measure of performance. Target detection performanceand absenteeism should have been treated as two separate measures. No analyses were even

13

provided to indicate the two sets of measures are related. For system performance (safety), onlyone of these classes of measures is actually critical.

A close analysis of the data yields some interesting findings. The coefficient of correlationbetween APSS/PI and the total testing errors composite score is only r = +.163. This accountsfor only 2.6 percent of the variance. An examination of individual data indicates that within thegroup classified as "not recommended for hire", based on the APSS/PI (n = 18), 74 percent of thetesting errors within -that group are made by only 33 percent (n = 6) of that subgroup. Mosterrors were apparently made by only a few individuals. Four of these "not recommended"screeners accounted for 18 of the 27 errors (67 percent). The performance of these fourindividuals (the only screeners with more than two errors) is clearly an anomaly. In fact, 12individuals in this subgroup made one or no testing errors.

Eighteen percent (18 percent) of the "recommended" group were also reported to have at leasttwo screening errors. Given the comparatively small sample sizes and substantial individualvariances it is not likely these data can be replicated in cross-validation. More importantly, theAPSS/PI as used to separate screeners into "recommended" versus "not recommended"subgroups was ineffective in classifying screeners by error rates. The contractor offers norationale for this inconsistency. The claim that APSS/PI accurately predicts objectiveperformance is completely unsubstantiated by the data offered. An alternative analysis shouldhave separated screeners into groups based on performance (errors), not APSS/PI scores. A "notrecommended" group could have represented all screeners with two or more errors while a"recommended" group could have been composed of all screeners with no screening errors.Comparison of these groups with regard to APSS/PI scores would have generated moreinterpretable results.

In fact, using the APSS/PI to make hiring decisions would have eliminated 40 percent of thosenot recommended for hire who had error free performance while employing a third of therecommended sample who had performance errors. The relationship between the overallnegative performance composite score and APSS/PI (r = +.16) indicates little improvement inprediction of satisfactory performance by incorporating job tardiness as a measure ofperformance. Overall, the data do not justify use of the APSS/PI in a selection environment. Theselection algorithm clearly has an unusually high false positive rate and is based on a weakrelationship using very small sample sizes.

Perhaps the most confusing aspect of the data is the relationship between supervisory ratings andthe APSS/PI. The APSS/PI does not predict supervisory performance evaluations. Theinvestigators simply explained that supervisory perceptions are not an accurate reflection ofobjective job performance. While this may be true, no data are presented that examine therelationship between testing errors and supervisor evaluations of performance although these datawere available to conduct such an analysis. An alternative explanation could be that supervisoryevaluations were an accurate prediction of objective job performance but the APSS/PI is notsensitive enough to predict these differences. It is never clear why the contractor completelyrejected the use of the subjective ratings to validate the screener's job performance.

14

A final issue is with regard to the effect of experience on job performance. Although the datawere available, no comparison was made between months on the job and any testing errorsmeasures. In a selection and training system it is important to understand the effects ofexperience or training on actual job performance. It is plausible that after several months ofexperience as a screener, performance would improve such that errors in target detection becomean unlikely occurrence. The contractor studies never examined experience as a possible factor.

While these investigations demonstrated the need and requirement for effective personnelselection practices, they do not offer a valid, reliable, and comprehensive instrument that can beused across the aviation security industry. Perhaps the industry's overall rejection ofimplementing this tool is the best indication of its inadequacies.

Although the contractor was the only research-based effort that attempted to identifyunsuccessful screeners prior to employment by using a personality trait model, other efforts havesought to distinguish successful from unsuccessful airline screeners using demographiccharacteristics. However, current statutes regarding hiring practices have negated the use ofmuch of this work in a selection environment.

3.3.2.2 International Total Services (ITS) Study.

One aviation security contractor recently sought to determine if there is a relationship betweenscreener performance and specific employee characteristics. In 1993, the Training and PersonnelDevelopment Department at International Total Services (ITS) completed a study ofdemographic factors and their relationship to job longevity and performance on FAA checkpointtesting (ITS, 1993). ITS compared these performance criteria for a sample of 3,183 screenersdivided into sub-groups by: (a) age, (b) sex, (c) ethnic background, (d) educational level, (e)previous employment, (f) military background, and (g) citizenship. Although these results cannot be used for employee selection and hiring, some findings are nevertheless worthy ofmentioning.

In a comparison of screeners with three or more years tenure versus the total sample, severalsignificant differences were identified. When compared to the nationwide screener population,high tenured screeners were on the average, older, better educated, more likely to be female,more likely from Asian than black ethnic background, and less likely to possess U.S. citizenship(see table 1). Nearly one-third of the high tenured sub-group were of Asian background (nearlyequally divided by gender), with another 17 percent Caucasian females. Employment tenure wasalso found to be twice the duration for the "housewife" subgroup than any other prior workexperience sub-group. Perhaps the most surprising finding however is the relationship betweeneducational level and job tenure. ITS reports that screeners with four year college degrees andthose with eight years of college have the two highest employment duration averages,respectively. This probably reflects the current poor outlook in the job market for collegegraduates and a work environment that is attractive for retired professionals. Interestingly,demographic factors were not useful in differentiating screeners who remained employed lessthan 60 days from the total sample.

15

TABLE 1. COMPARISON OF SCREENERS NATIONWIDE TO VETERAN SCREENERS(DATA FROM ITS PERSONNEL PERFORMANCE STUDY, 1993, SECTION III:

EMPLOYEE RETENTION, GRAPH 1)

TOTAL SAMPLE SCREENERS WITH > 3 YEARSOF SCREENERS (N = 3,183) WORK EXPERIENCE (N =

Average age 38 years 54 yearsFemale 45.4 percent 57.2 percentAsian 10.9 percent 30.9 percentBlack 32.4 percent 12.2 percentU.S. Citizen 82.9 percent 63.0 percent

* The sample size of the veteran screeners was not reported.

Of potentially great importance is the ability to identify screeners who do not performeffectively (i.e., fail FAA checkpoint tests). The data presented by the ITS study however, donot demonstrate a relationship between demographic factors and the number of FAA testfailures over the preceding 12-month period. The only substantial difference found is that theblack subgroup is more likely to fail FAA tests than any other ethnic group.

Similar to findings for the job tenure criteria, several demographic factors distinguish superiorperformance in screeners (i.e., as determined by security firm evaluations) from the totalnationwide sample. A group of 159 screeners were selected for special recognition because they"... showed an exceptional ability to repeatedly pass FAA tests." This selected group differedfrom the overall nationwide sample in: (a) mean age (51 versus 38 years); (b) gender (54 percentversus 45 percent female); 3) ethnic background (50 percent versus 24.2 percent Caucasian); (d)FAA pre-employment average test scores (98 versus 93.5); and (e) experience background(nearly a three-fold difference in prior military background and twice the percentage of the totalsample with a law enforcement background).

As previously stated, the use of these data in making hiring decisions is unacceptable undernumerous federal and state statutes. From a methodological standpoint, there are a number ofcritical issues regarding the data that need to be addressed that impact the interpretation of thefindings as well.

All the ITS data are presented in actuarial format using frequency tabulations as the primarymeans for the analysis. There were no statistical methods applied to the data to assess either themagnitude of the differences shown, or to determine if statistical significance existed. Non-parametric techniques (e.g., Chi-Square analyses) could have been used on nearly allthe data andwould have provided a more meaningful interpretation.

A second critical issue concerns the methods used to sub-divide the sample population. In mostcases, the data are simply grouped along one dimension (e.g., gender) rather than multipledimensions (e.g., gender X age X ethnic background). Consequently, assignment to a subgroupis severely confounded. For example, the sub-groupings within the age grouping can be

16

confounded by educational level by a high proportion of retired professionals. Similarly,citizenship assignment could conceivably be compromised by one specific ethnic group. Withoutconsidering the multi-dimensional nature of the demographic factors involved, it is difficult toisolate the effect of specific factors.

Most importantly, the data are only partially useful in identifying superior performers (i.e., asassessed by company evaluations) or job longevity post facto. Regardless, these factors cannot

be used for promotion decisions or incorporated into employee recognition programs. Althoughnot conclusive, the findings offer no evidence for predicting which screeners have a propensityfor committing screening errors. In addition, the results are not useful in identifying the newapplicant who will leave the career field within the first two months. A strong selection systemmust have the potential of identifying the short-term or ineffective applicant before employment.

3.4 AIRLINE PASSENGER SECURITY SCREENER TRAINING.

A well-trained staff is key to providing an effective airline security program. Most of theemphasis however, has been placed on developing technology (e.g., neutron techniques, gammaray techniques) and very little has been directed toward enhancing human capabilities.

Most screeners are trained with a limited inventory of obvious test items and are only taught torecognize complete items-not their individual components. Some security firms though, werefound to exceed the required training minimums. These firms have developed their owncirriculum to supplement the FAA/airline requirements as well.

3.4.1 Syllabus Topics.

Currently, most security companies provide a three-part training program for airline securityscreeners. Part one incorporates 12 hours of classroom training over a three day period (see table2. Part two involves 40 hours or supervised on-job-training (OJT), while part three entailsrecurrent training scheduled after a six-month period.

In order to satisfy the classroom training requirements, the ATA has developed the "AirlineSecurity Screener Training" manual, an instructor's guide approved by the FAA for use incheckpoint security training. The manual includes 13 lesson plans in order to provide guidanceto the instructors and ensure a comprehensive course of instruction. Table 2 below brieflyillustrates each lesson and the allotted time.

17

TABLE 2. ATA-APPROVED CLASSROOM SYLLABUS FORCHECKPOINT SECURITY SCREENERS

LESSON DURATIONPLAN TOPIC(S) (MINUTES)

Session 1 The First Line of Defense (video) 10The Reason for Passenger Screening 10The Legal Basis for Passenger Screening 15The Effectiveness of Passenger Screening 15

Session 2 How Aviation Security Works 50Screening Techniques (video)

Session 3 The Importance of the Screener 50Screening Techniques (video)

Session 4 Identifying the Threat 50Screening Techniques (video)

Session 5 Identifying the Threat 50Session 6 Screening the Passenger 50

Screening Techniques (video)Session 7 Screening the Passenger 50

Screening Techniques (video)Session 8 Screening Hand-Carried Items 50

Screening Techniques (video)Session 9 Screening Hand-Carried Items 50

Screening Techniques (video)X-Ray Image Interpretation Teaching Aid (video)

Session 10 Screening Atypical Passengers 50Screening Techniques (video)

Session 11 Special Situations 50Session 12 Security Information Circulars and Security Directives 25

Denied Entry into the Sterile Area 25Session 13 Guidelines for Station Specific, Instruction as necessary

It's Your Responsibility (video)

The Transport Canada's Security Training and Development Division has also developed a"Screening Personnel Course" training manual for security screeners in Canadian airports. Themanual is currently being revised to include new training procedures and policies. To comparethe two training programs, the following topics from Transport Canada's manual are illustratedbelow:

a. Introduction to Screening (job familiarization, duties, responsibilities, andauthority)

18

b. Weapons, Explosives, Dangerous Articles and Dangerous Substances(types/kinds of threats, threat identification, how to react, who to contact);

c. The Walk Through Metal Detector (types of metal detection units,operational features, functional guidelines);

d. The Hand Held Metal Detector (purpose, familiarization, limitations,operational procedures, maintenance);

e. Body Search of Persons (purpose, applicability, guidelines andprocedures);

f. Scanray Linescan® System 1 and 2 (familiarization, safety precautions,operational procedures, law enforcement assistance);

g. Hand Search of Baggage (when required, baggage requiring specialconsideration, exceptions, guidelines and procedures);

h. Special Situations (diplomatic missions, aircraft crews, handicappedpersons, persons escorting prisoners, etc.);

i. Terrorism (types of terrorism, terrorist incidents, terrorist profile andcommon characteristics).

The OJT phase (part two) is typically not formalized, with little standardization betweencompanies. However, the OJT environment is really where screeners develop the criticalexpertise necessary to perform their job successfully. There is even less emphasis placed uponthe importance of recurrency training (part three) in many companies. This phase of training isoften based upon a review of the ATA training tapes illustrated in the beginning of this section.In consideration of the high turnover rates in this industry, it is obvious that a significantpercentage of screeners never reach this final phase of training.

3.4.2 Research on Training to Improve Screener Performance.

An early study (Potter, 1975), as reported in Funke (1979), examined the ability of perceptualtraining techniques on developing the skill of detecting weapons for screeners. Data wereobtained from 18 subjects, of which one-half were college students. All were high schoolgraduates. Subjects were randomly assigned to one of three groups: Group 1 was trained withsound/slide presentations dealing with explosives; Group 2 was trained with an audio/visualprogram on guns; and Group 3 (the control group) was trained with written instructionsconcerning items they should watch for during visual inspections. Subjects were shown 95 slidesof X-rayed baggage of which approximately 30 percent contained various weapons. Each slidewas shown for 6 seconds followed by a 4 second blank period. Subjects were to indicate whetherthey thought the bag required opening, and if so, the reason for their decision.

19

Three measures of performance were analyzed: (a) number of bags containing weapons whichwere not found (i.e., missed detections); (b) number of bags which were incorrectly identified ascontaining weapons (i.e., false alarms); and (c) number of bags correctly identified as containingweapons but the weapons themselves misidentified. The number of false alarms and the numberof misidentified weapons were not significantly related to the training conditions. However, thetype of training had a significant effect on the number of weapons found, although no statisticaldata were reported by Funke (1979). The mean percentage of weapons missed for the threegroups were reported as follows: explosive trained (M = 10.0), gun trained (M = 22.6), andcontrol group (M = 32.8). Funke indicated that these findings suggest that perceptual trainingtechniques (e.g., sound/slide presentations) can significantly improve target identification onvisual inspection tasks, particularly when targets appear infrequently.

Another study by Nadler and Mengert (1993), investigated the benefits of computer-basedinstruction (CBI) compared to current methods of selection, training, and screener certification.They used the Safe PassageTM System, a CBI system developed specifically for airline passengersecurity screeners. The Safe PassageTM System presents screeners with X-rayed baggage imagesstored in a video database of approximately 2,000 images. The images represent eightcategories: (a) innocent, (b) suspicious innocent, (c) electronic innocent, (d) explosive, (e) gun,(f) knife, (g) other sharp objects, and (h) combined/other weapons.

Nadler and Mengert (1993) were given data on 1,465 screeners who worked for securitycompanies at five major domestic airports equipped with the Safe PassageTM System. Thepreliminary findings, as reported in their interim report, are based on data from 500 screenersreferred to as their Sampled Data Set. Screeners voluntarily performed "simulated" screeningtests while threat images were presented randomly. Each test contained 12 different images.Three levels of proficiency were programmed (i.e., low, medium, high) so that each levelcontained increasing percentages of relatively "difficult" test images: 25, 50, and 75 percent,respectively.

They found that four of the eight image categories (i.e., suspicious innocent, explosive, knife,other sharp object) showed accuracy less than 80 percent in the Low Proficiency Level, reportingaccuracy levels of 75.1 percent, 77.4 percent, 76.8 percent, and 77.3 percent, respectively. Thus,images that fall into these categories can generally be more difficult to detect. Preliminaryresults also indicate a low percentage of "critical errors" (i.e., errors resulting when a screenerpasses a bag that should have been held for further inspection). They noted that 50 percentcommitted no critical errors at each proficiency level. Based upon these initial data, Nadler andMengert (1993) concluded that there is, however, ample room for improvement in screener'sability to distinguish threat items from innocent items in X-ray images of baggage. This CBIsystem appears to be the most effective and advanced training program available on the markettoday.

In the most recent study to date on airline screeners, Kaempf, Klinger, and Wolf (1994)employed a cognitive systems engineering approach (CSE) to identify the decision requirementsof security personnel using X-ray scanning devices at airport security checkpoints. Their primaryobjective was to develop decision-centered interventions that will enhance screener performance

20

either through training or through the implementation of decision supports (e.g., patternrecognition system). Forty-three security personnel were interviewed, including managers,trainers, screeners-in-charge, and checkpoint screeners.

"Screening companies must recruit individuals who have no applicable skills for minimumwages and instill some measure of proficiency in a relatively short period of time" (Kaempf etal., 1994, p. 29). They also identified several serious problem areas with current training

programs. First, none of the existing training programs address the cognitive skills required forscreeners to perform their jobs proficiently. Second, the OJT program is not a well-structured,controlled effort designed to enhance the skills of the novice screener. Finally, recurrent trainingis no more than a review of training received during initial hiring and is similarly ineffective.

Phase I of this study (Kaempf et al., 1994) has yielded information highlighting the differencesbetween novice and expert screeners in their decision-making processes. Their findingsindicated that experts have developed, through accrued experience, an extensive mental "library"of both non-threat and target (threat) items. This mental "library" facilitates the screener's abilityto rapidly and accurately differentiate between the preponderance of non-threat items; and theinfrequent threat items-a critical skill. Kaempf et al. stated, "What is needed is an integratedtraining program that considers the skills required to perform proficiently and that uses trainingopportunities to build on skills previously developed" (p. 30).

3.5 SCREENER TURNOVER RATES.

Most aviation security firms experience high turnover rates-in some cases, 100 percent in just10 months. In their recent study on airline screeners, Kaempf et al. (1994) identified severalfactors which lead to such high turnover of personnel, including low pay, lack of opportunitiesfor advancement, and competition for these types of workers in the job market. Kaempf et al.stated, "The facilities that we visited typically have a stable employee base of 10 percent with theremainder in constant flux. This results in about 90 percent of all screeners at any givencheckpoint having less than six months experience" (p. 29). High turnover rates such as thesewill ensure that very few screeners will become highly experienced.

The extremely high turnover rates so prevalent in the industry for the airline security screenerposition have two critical implications. First, it inhibits the development of expertise; andsecond, it reduces the ability to take advantage of an effective and cost-efficient method oftraining (i.e., OJT from experienced screener "mentors"). "The constant migration of personnelcreates significant requirements for training that strain available resources. For example, there isa continuous need to conduct initial training for new hires" (Kaempf et al., 1994, p. 29).

An examination of the starting wages offered to new hires presents some possible evidence forsuch high turnover rates (see table 3). As a general rule, occupations with such highresponsibility normally necessitate higher levels of compensation and benefits.

21

TABLE 3. STARTING WAGES AND BENEFITS FOR AIRLINE PASSENGER SECURITYSCREENERS BY AIRPORT AND SECURITY CONTRACTOR.

HOURLYBENEFITS LOCATION COMPANY WAGE

JFK Airport Argenbright $4.50Yes

Newark Airport ITS $5.05No

O'Hare Airport Andy Frein $4.25No

Orlando Airport Argenbright $4.25No

3.5.1 Incentives/Reward Systems.

Several years ago, Hertzberg (1966) noted that achievement, recognition, and responsibility arethe most consistent motivational factors producing job satisfaction. Many theories state thatgood job performance leads to job satisfaction rather than job satisfaction leading to good jobperformance (Brennan, 1974). Wallis (1992) reported that reward systems generally stimulateattention and have proven effective in many airport security checkpoints. An effective incentiveprogram can also lead to significant gains in productivity.

Wiener (1975) argues that an offer of a 'substantial financial reward' does not significantlyinfluence accuracy in inspection tasks. According to Wiener (1975), Murrell (1965) explains,"It must be realized that inspection is largely a perceptual task and ... successful perception is nota process which can be readily influenced by an incentive, however great" (p. 110). Furthermore,Wiener (1975) noted that financial incentives in an experimental setting have generally yieldedunimpressive results in inspection studies.

3.6 VIGILANCE OPERATIONAL SIGNIFICANCE FOR AIRLINE PASSENGERSECURITY SCREENERS.

For nearly 40 years, vigilance researchers have engaged in one of the most prolific programs ofscientific research in any area of behavioral science. Well over a thousand studies have beenpublished during this time (Mackie, 1987). Over 30 years ago, Buckner and McGrath (1963)conducted a symposium on vigilance and addressed the following questions: "Why do peoplefail to detect important signals? Under what circumstances do they fail to detect them? Whatcan we do to ensure that they detect them?" (p. vii). This section of the report will attempt toexplore these questions as they relate to the airline passenger security screener occupation.

Human vigilance is the ability to detect rare signals over a prolonged period of time. Simply put,it is the ability to maintain attention. Vigilance has been described in several ways: asperformance on monitoring tasks; as attention over extended periods of time; as a state of

22

maximum physiological efficiency; and as a state of readiness to respond to infrequent, yet smallchanges occurring at unpredictable random intervals (Buckner and McGrath, 1963; Davies, 1979;Macworth, 1957). Vigilance is synonymous with long-term attentive behavior (Davies andParasuraman, 1982) and sustained attention (Parasuraman, 1986). According to Jerison andWing (1963), vigilance is a general problem of attention or alertness. The word attentionhowever, unlike vigilance, is used with little confusion in daily affairs (Edwards, 1990). Otherterms used in the literature include monitoring and search-all are treated as aspects of attention(Parasuraman, 1986). The authors will not make any distinction between vigilance, attention,monitoring or search in this report. The reader is encouraged to review Parasuraman (1986) forcomplete definitions and key concepts and terms.

Vigilance performance is required in a number of industrial settings. To perform successfully,operators must remain alert for indications of malfunctions and changes in operating states. Forexample, the air traffic controller who keeps aircraft separated by observing 'blips' on a monitor;the flight engineer who observes various engine performance parameters (e.g., oil temperature,oil pressure, fuel flow) to ensure a safe flight; the quality control inspector who examines anumber of products (e.g., circuit boards, microchips, beverages) to detect and remove deflectiveor flawed items; and the anesthesiologist, who monitors a patient during an operation to ensureall vitals signs are within an acceptable range (Howland and Wiener, 1963; Wickens, 1992). Ourdefense against a nuclear attack depends ultimately upon the vigilance of the persons observingthe displays in our early-warning stations throughout the world (Buckner and McGrath, 1963).

3.6.1 Origins of Vigilance Research.

Research on vigilance, monitoring, and search arose in response to problems identified duringWorld War II. In 1943, the Royal Air Force (RAF) asked if experiments could be conducted todetermine the optimum length of watch for radar operators on anti-submarine patrol. RAFreports documented that a number of potential U-boat contacts were being missed and that theoperators were overstrained (Davies, 1979; Davies and Parasuraman, 1982; Parasuraman, 1986).After some preliminary experiments conducted by Mackworth (1950), the RAF began anoperational study of the detection of submarines by radar operators. They found that after 30minutes on watch, a marked deterioration in performance occurred-thus coined the term"vigilance decrement." This performance decrement has been a recurrent finding in countlesslaboratory studies observing human behavior and will be discussed in a subsequent section ofthis report.

3.6.2 Theories of Vigilance.

There have been a number of theories advanced over the years that attempt to explain thevigilance phenomenon (e.g., arousal theory, expectancy theory, inhibition theory, signal detectiontheory). Although no particular theory is likely to be entirely correct, Wickens (1992) noted thatthe advantage of such theories is that they provide ways of accounting for vigilance loss andthereby suggest techniques to improve vigilance performance.

23

Two of the more popular vigilance theories that have applications for airline passenger securityscreeners are the expectancy theory and the filter theory. According to Davies and Parasuraman(1982), the expectancy theory hypothesizes that the observer develops expectancies about thefuture probability of a signal occurrence on the basis of previous experience(s) with the task, andthat the detection rate is determined by an expectancy level. For example, if the probability ofoccurrence is low, as it is for screeners, then the expectancy for observing a signal is low. Basedon this theory, Wickens (1992) noted that introducing false signals will help keep the screenerattentive, particularly if the signals are physically similar to the real signals themselves. Thus,according to the expectancy theory, the periodic tests carried out by the FAA and airlines notonly allow for a measure of performance, but also helps keep screeners on their best possiblevigilant behavior.

For the airline passenger security screener to maintain a 100 percent detection efficiency usingthe expectancy theory, the screener must assume that every bag and every passenger contain orpossess a weapon of some sort. Although realistically, 100 percent detection efficiency seemsvirtually impossible, some simple techniques have shown signs of progress. For example,Wickens (1992) reported that simple instructions describing new policies and/or procedures candramatically improve detection performance. Wickens stated, "... in airlines security inspection,increased stress on the seriousness of misses (failing to detect a weapon smuggled through theinspection line) caused a substantial decrease in the number of misses from 1987 to 1988" (p.46). The source(s) of these data however, were not reported by Wickens.

The second theory that has implications for the airport screener is the filter theory. Developed byBroadbent in the 1950s, the filter theory is based on experiments conducted in the areas ofselective attention, vigilance, and the effects of noise on performance (Davies and Parasuraman,1982). The filter theory proposes that there are periodic failures on the part of the operator totake in task-relevant information and the net result is a decrease in the number of detections (i.e.,missed targets) and an increase in reaction times (i.e., longer time to react). Loeb and Alluisi(1984) note that this performance decrement is particularly marked when operators aremonitoring channels of information where there is considerable repetition (e.g., screening forweapons). According to Davies and Tune (1970), the temporary failure to take in task-relevantinformation is attributed to the presence of competing stimuli, since different classes of stimuli(e.g., size, shape, complexity, novelty) are competing with one another to enter the filter. Thefilter can however, be trained to select certain types of information (e.g., guns, knives, dynamitesticks) by properly instructing the operator through repeated practice or test sessions.

The filter theory assumes that the operator's information-handling capacity is limited and thathe/she cannot analyze more than one stimulus simultaneously, therefore competing stimuli areselected sequentially for analysis (Loeb and Alluisi, 1984). The selection of each stimulus andthe order in which they are analyzed depends on certain stimulus features, stimulus importance,and novelty of the stimulus (Broadbent, 1958).

24

3.6.3 Vigilance Decrement.

In some vigilance situations, although not all, the number of signals correctly detected (i.e.,detection or hit rate) has been found to decline with time on task (Davies and Tune, 1970). Fox(1975) reported that with prolonged inspection periods, detection performance can deterioraterapidly, and "... drops of 40 percent in 30 minutes have been noted" (p. 89). The magnitude ofthe vigilance decrement in various studies has been inconsistent however, with some studiesreporting miss rates as high as 40 percent (Craig, 1984).

Since the early efforts, research on vigilance has primarily been concerned with reaching anunderstanding of the factors which are responsible for this decrement in performance (Davies andParasuraman, 1982). The vigilance decrement is simply the inability to maintain vigilantbehavior (i.e., attention or alertness) over extended periods of time (Pierce et al., 1991). Thisperformance characteristic has been a focal point of the "critics challenge" and is generallyuncritically accepted, even though some investigators (Smith and Lucaccini, 1969; Wiener, 1975;Wylie, Mackie, and Smith, 1985) seriously doubt its existence in real-world settings. Theycontend that the vigilance decrement only occurs in laboratory settings and not in the real world,and is merely a result of fading motivation of the subjects. Others (Nachreiner, 1977; Wiener,1987) support the same viewpoint. Wiener (1987) stated: "There is no evidence ... to suggestthat vigilance decrements exist in real-world systems and, indeed that it is anything but alaboratory artifact brought on by contrived experimental tasks, unmotivated subjects, and poorinstructions" (p. 730). Furthermore, Davies and Tune (1970) noted, "... under such conditions[i.e., laboratory experiments], it is not surprising that observers sometimes fall asleep and thatreports of drowsiness and extreme boredom are common" (p. 11).

There has not been an experiment to date that either confirms or denies the existence of thevigilance decrement (Wiener, 1987). While there are ample studies suggesting that the vigilancedecrement does in fact exist, some field studies have yet to encounter this problem. For example,Nachreiner (1977) reported on the performance of coin inspectors over several days and they didnot reveal any decrement with time on task.

There have been a number of theories advanced that attempt to explain the vigilance decrement(e.g., arousal theory, habituation theory, motivation theory). Two of the more popular theoriesare the arousal theory (or activation theory) and the motivation theory. According to Davies andParasuraman (1982), the arousal theory maintains that a progressive reduction in the arousallevel of the central nervous system takes place during task performance. This decrease in brainactivity is largely brought about by the monotonous nature of the vigilance situation, and as aresult, the brain becomes less responsive to and less efficient at dealing with external stimulation.On the other hand, the motivation theory maintains that the vigilance decrement is attributable toindividual differences in motivational level (e.g., some people being more conscientiousmonitors than others), as well as to reductions in motivational level caused by the monotonousconditions of work and the failure to provide adequate incentives. It is frequently viewed asbeing of crucial importance in the debate about the practical significance of vigilance research(Davies and Parasuraman).

25

Other investigators (Pierce et al., 1991) purport that the vigilance decrement depends upon thetask-related variables (e.g., frequency or density of monitoring, display size, display brightness,viewing angle, viewing distance) and the individual characteristics (e.g., age, intelligence,motivation, personality). More on how these characteristics can affect vigilance performance isreported in a subsequent section of this report.

Nachreiner (1977) presented a different viewpoint regarding the potential reasons for thisapparent performance decrement. Based on his findings from an optical vigilance experiment,Nachreiner reported that performance decrements in vigilance experiments are dependent on thesubject's perception and realization of the experimental situation. Nachreiner recruited subjectsby two announcements: one asking for participation in a psychology experiment, and the otherasking for applicants for the job of experimenter. Eight subjects were equally divided into twogroups, the control group and the experiment group. The subjects had to distinguish a series ofspikes simulating bioelectrical potentials where the critical signal had 10 spikes and the neutralsignal had eight. Nachreiner found a decrement after the first 30 minutes for the experimentgroup, but not for the control group. These findings suggest that the vigilance decrement mightbe more dependent on the subject's perception of the experimental condition rather than on thespecific task characteristics of the experiment conducted. These findings however, should beviewed as inconclusive because of the limited number of subjects.

3.6.4 Vigilance Tasks.

Most vigilance tasks entail observing a display and monitoring it for very subtle or obviouschanges in state. Davies (1979) stated that vigilance tasks are tasks "... in which attention isdirected to one information display ... over long, unbroken, periods of time, to detect infrequentchanges in the state of the display that are extremely difficult to discriminate" (p. 14). Otherinvestigators (Buckner and McGrath, 1963; Davies and Parasuraman, 1982; Edwards, 1990) referto vigilance tasks as "monitoring" or "watch keeping" tasks.

Craig and Colquhoun (1977) stated that a major criticism surrounding vigilance research is thatmost tasks used in laboratory studies are too simple to have any relevance for real-life operations.Nachreiner (1977) stated, "There are many differences between vigilance experiments in thelaboratory and monitoring jobs in field situations. Because of these differences, the question hasbeen raised whether vigilance experiments are relevant to the problems of monitoring in fieldsituations" (p. 666). Wiener (1987) stated: "... the nature of the monitoring task, which operatesin a real-time, unpredictable, event-driven environment for long periods day after day, it isunlikely that any laboratory experiment could be set up even to faithfully simulate, let aloneduplicate, a working system" (p. 727).

Numerous investigators have examined the effects of vigilance on simple and complex tasks.However, most studies were difficult to compare because they employed different kinds ofvigilance tasks, different response types, and different response measures (e.g., hit rate, reactiontime, false alarms). Wiener (1987) stated: "The 'monitor' actually time-shares betweenmonitoring and more active tasks [e.g., searching suspicious bags], and thus the job is not aspassive, routine, or cognitively unenlightening as the papers on vigilance would imply" (p. 730).

26

Mackie (1987) stated: "Countless variables have been identified that may influence thedecrement or level of performance" (p. 707).

Table 4 presents 15 of these variables that influence vigilance performance. Many of thesereported variables do not represent an immediate threat to airline passenger security screeners.For example, performance decrements associated with task duration are often countered byrequiring screeners to rotate positions every 20-30 minutes. This rotation reduces the probabilityof error(s) associated with the task monotony. In addition, improving employee recognitionwould work as an incentive by boosting motivation and morale among employees.Implementation of such a program could be quite successful, as well as very cost-effective. Otherfactors however, (e.g., circadian rhythms, payoff and/or rewards, motivation, morale, socialenvironment), present some very obvious challenges to the screener occupation. Very little isknown how these factors would impact screener performance; thus, further research is needed toaddress these aspects of screener tasks.

TABLE 4. PARTIAL LISTING OF FACTORS THAT AFFECT PERFORMANCE INVIGILANCE TASKS (MODIFIED FROM MACKIE, 1987, P. 708)

Task durationWork/rest patternCircadian rhythms (first shift versus third shift)Sleep loss; sleep qualityIncentives; payoff and/or rewardsMotivation (intrinsic, extrinsic)Display design (rate of presentation, display size, lighting/illumination)Individual characteristics (age, gender, intelligence, experience, intro-extrovert)Mood or moraleEnvironmental stressors (noise, sound, vibration, lighting, temperature)Social environment (peer pressure, supervisor/management pressure)Illness; injuryDrug use (caffeine, nicotine)Exercise, physical work

3.6.5 Performance Measures in Vigilance-Related Tasks.

Performance measures taken from vigilance tasks are usually a measure of operator efficiency.Most measures are generally assessed by one of three ways: (a) the number of correct detections(known as the detection rate or hit rate); (b) when a signal has been reported and none has beenpresented (known as false alarms or false detections); and 3) when a signal has been presentedbut not identified (known as missed detections). A fourth measure, but not as common, is theamount of time taken to detect a signal, referred to as the detection latency (Davies andParasuraman, 1982; Davies and Tune, 1970).

27

The most commonly used measure of performance in vigilance studies is the detection rate(Davies and Tune, 1970); however, much useful information can be obtained from the otherperformance measures. According to Davies and Parasuraman (1982), the primary measures ofperformance in vigilance situations are the detection rate, the false alarm rate, and the detectionlatency. All three measures are essential for understanding the way in which vigilanceperformance varies with time on task, across different experimental conditions, and betweendifferent individuals.

In an experimental setting, performance measures for airline screeners could include: (a) thedetection rate (i.e., number of bags correctly identified as containing weapons); (b) false alarms(i.e., number of bags incorrectly identified as containing weapons); and 3) missed detections (i.e.,the number of bags containing weapons which were not found). This method was employed byPotter (1975) as previously reported on page 31.

In industry settings however, measuring vigilance performance is difficult to say the least. Forexample, the number of missed detections (i.e., weapons that make it past the screeningcheckpoint) is not known; therefore, it is difficult to assess this particular aspect of performance.Most of the tests used to measure vigilance performance in industry settings are validated againstsupervisors' ratings (Wiener, 1975). This subjective criteria is questionable however, since thesupervisor may be totally unaware of the actual performance of the individual inspectors. In lightof this issue, Wiener stated, "His ratings [the supervisor's] are probably based on perceptions ofearnestness and cooperation, and the correlation between these attributes and actual inspectionperformance is unknown" (p. 102).

Despite the inherent limitations in measuring vigilance performance in the actual workingenvironment, by using test weapons such as those currently employed by the FAA and aircarriers, we could at least identify those screeners who fall below the accepted standards. Onceidentified, these screeners could then be sent for additional training to focus on their area(s) ofweakness, thus theoretically, improving the overall efficiency of the security system.Nonetheless, the FAA and airline "tests" lack the ability to accurately determine who the betterperformers are because of their apparent and repeated use of similar test objects.

3.6.6 Techniques to Combat the Vigilance Decrement.

There are some methods and/or techniques that can increase vigilance performance. Wickens(1992) noted that any technique that will enhance the subject's memory of the signalcharacteristics (e.g., object shape, size, density) should reduce performance decrements andpreserve a higher level of awareness. For example, by introducing false signals (e.g., testweapons), one improves or heightens the operator's awareness, consequently improving thesystem's effectiveness. Furthermore, Kelly (1955) as reported in Wickens, found a large increasein detection performance when quality control operators were allowed to look at televisionpictures of idealized target stimuli. This emphasizes the importance and continued practice ofusing FAA and airline test objects to continually refresh the screener's memory and heightentheir awareness, despite the objects' obvious shortcomings.

28

Another method to combat the vigilance decrement was proposed by Childs (1976). Herecommended that inspectors should have access to visual representations of possible defectivesrather than the representation of those that are normal. This position lends further support to theexpansion of the FAA and airline test items to include "parts" or "components" of a threat, ratherthan the complete and assembled item. Other investigators (Davies and Parasuraman, 1982;Craig, 1984) have also suggested different methods for reducing the vigilance decrement. Mostsuggestions are extrapolations from laboratory experiments; they include:

a. provide appropriate work/rest cycles;

b. instill motivation by emphasizing task importance;

c. introduce mild environmental stress (heat, noise, vibration);

d. provide observation by supervisors;

e. use personnel selection techniques to identify individuals with a propensity formaintaining vigilance.

3.7 SELECTION TEST DEVELOPMENT FOR U.S. ARMY PERSONNEL.

Very few laboratory studies have involved watch durations or performance tasks that evenremotely resemble airline passenger security screeners. However, of particular interest is therecent work of Crumley, Pierce, Schwalm, Coke, and Brown (1992). They conducted anexperiment with U.S. Army enlisted personnel to determine if cognitive style tests could be usedto predict target detection performance. Such tests, if validated, could prove invaluable to theairport security industry while screening for good performers during the pre-employment process.The purpose of their research was to determine if scores from a group of cognitive factors (i.e.,

Speed of Closure, Flexibility of Closure, Perceptual Speed) and tests (i.e., Gestalt Completion,Concealed Words, Snowy Pictures, Hidden Figures, Hidden Patterns, Identical Pictures), orselected biographic items (i.e., gender, smoking habits, coffee drinking) predict the ability ofpersons to detect targets. They reported that the Speed of Closure and the Flexibility of Closurefactors were used because they appear to "... involve factors important to the task of seeingpartially obscured items" (p. 6). They further noted that the Gestalt Completion Test was usedbecause its items appear to be almost identical to the visual and intellectual tasks involved indetecting camouflaged targets where portions of the items blend into the background.

Crumley et al. (1992) presented 1,440 slides developed from photos taken with a 35mm camerafrom a helicopter flying between 400 and 800 feet over and near Fort Still, Oklahoma. The slidesdepicted aerial views of terrain where subjects were to identify man-made objects other thanroads, railroads, power lines, and fences (e.g., vehicles, bridges). Of those 1440 slides, 90 (6.3percent) had actual targets. The slides were presented on a 19" black-and-white televisionscreen. Data from 209 subjects were used in the analysis. Two measures of performance wereused in their experiment: (a) targets detected, and (b) false detections.

29

Their findings indicate that certain biographical items can be found to affect a person's ability todetect targets. They found that persons who fish for amusement make more target detectionsthan those who do not fish (F = 3.30, p = .039). Their rationale for this finding is that fishing isoften appealing to persons who do not become restless in situations where attention must bemaintained over prolonged periods. They also found that people who do not wear glasses makemore detections than people who have glasses prescribed (F = 5.09, p = .025). They noted thatthis was probably a result of glasses not being worn when they should have been. Lastly, theyfound that persons who drink moderate or small amounts of coffee daily (i.e., less than 6 cups perday) make more detections than heavy coffee drinkers (i.e., those who drink more than 5 cups perday) (F = 4.34, p = .014). Stimulants (e.g., amphetamines, caffeine) in small to moderate dosescan improve detection performance by increasing the arousal level of subjects. This finding hasalso been reported by other investigators (Davies and Parasuraman, 1982; Mackworth, 1965;Wickens, 1992).

Based on their findings, Crumley et al. (1992) concluded that the Snowy Pictures test is perhapsthe best test for predicting target detection performance, and that it can be used in researchsituations where it is desirable to identify subject aptitude before the subjects are tested.Furthermore, they noted that the cognitive factor Speed of Closure and possibly the Flexibility ofClosure factor also predict target detection performance. They suggested that cognitive testscould form a basis for selecting persons needing additional training and development after theyare assigned to their particular occupational specialty. These findings appear relevant to airlinepassenger security screeners because of the task similarities (i.e., searching for and detectinghidden targets), however warrant further investigation.

3.8 EFFECTS OF STRESSORS ON SONAR OPERATORS.

Sonar operators play a critical role to military missions of the surface, sea, and air. The tasksinvolved in sonar operations are similar to those of airline passenger screening. For example,sonar operators must watch a display for extended period of time looking for small and subtlechanges in the display state. Wylie, Mackie, and Smith (1985) surveyed 212 Royal Navy sonaroperators (i.e., submarine operators, surface ship operators, helicopter operators) to gain a betterunderstanding of the stress conditions associated with the occupation and to identifycountermeasures that will help minimize those stress-related effects.

Most stressors are assumed to have a negative impact on operator performance, although thedegree of impact or degradation is unclear (Wylie et al., 1985). This uncertainty prompted Wylieet al. to collect data on the perceived impact and prevalence of stressors associated with sonaroperations. They divided stressors into four separate categories: (a) task-related stressors(boredom/monotony, operator workload, displays and controls, work station/personnelequipment design, midnight and morning watches); (b) environmental stressors (uncomfortableheat, uncomfortable cold, vibration, noise, uncomfortable air pressure, illumination or lightingproblems); (c) organismic stressors (fatigue, tiredness, minor illness); and (d) social stressors(command pressure). Note that no statistical data were presented in their report.

30

They found that there was a strong agreement among sonar operators concerning which stressorshad the most serious impact on operator performance. The two most common stressors reportedwere boredom and fatigue. Boredom, generally associated with monotonous operatingconditions, was viewed as the "worst" stressor impacting performance. The adverse impact(s) ofboredom is clearly evident in most monitoring occupations. The respondents reported thatboredom/monotony occurred very frequently (in fact, more so than any other stressor); therefore,it was perceived as more severe than any other stressor. Other investigators (Boff and Lincoln,1988; Thackray, Bailey, and Touchstone; 1977a) found similar detrimental effects of boredom(i.e., subjects that give self-reports of high boredom typically show longer reaction times andgreater performance decrements).

The impact of fatigue on operator effectiveness was also reported as a recurring problem. Fatiguewas ranked very high (i.e., strong detrimental effects) by the majority of the operators, and theimpact of fatigue was judged greatest on vigilance and overall operator effectiveness. Theyreported that fatigue was a result of both long work hours and poor sleep quality. The fatiguethey were concerned with however, was that of "mental" fatigue, and not due to prolongedphysical exertion. Others have reported on the effects of fatigue and their impact on taskperformance (Eysenck, 1983; Grandjean and Koji, 1971; Poulton, 1973).

These findings carry strong implications for airline passenger security screeners because of thesimilarity between tasks employed and the adverse consequences on operator performance. Giventhe perceived strong impact of boredom/monotony and fatigue, it would be desirable to develop alist of countermeasures that would help minimize these stress-related effects. This area appears tobe fruitful for research opportunities where the potential findings could be useful for numerousindustrial situations where monitoring plays a major role (e.g., nuclear power plants, air trafficcontrol, agriculture inspection).

4. HUMAN FACTORS ISSUES IN AIRLINE PASSENGER SECURITY SCREENINGCHECKPOINT WORKSTATION DESIGN.

The goals of this research program are to define the abilities and traits required of successfulairline passenger security screeners and to develop valid selection protocols for those abilitiesand traits. However, in the course of conducting the JTA it became readily apparent that severalhuman factors issues regarding workstation design needed to be addressed. Inadequacies inworkstation design can adversely impact screener performance and reduce the efficiency of theoverall security procedures. Poor work environments may also contribute to fatigue and mayaffect personnel retention. Supporting this concept, Astley and Fox (1975) noted that providingan improved working environment-as a result of physical changes (e.g., ergonomically designedchairs, padded floor mats)-will influence personnel turnover and absenteeism in a positivedirection. These issues are presented in brief since they are beyond the scope of the currentwork.

31

4.1 LIGHTING AND GLARE.

No standardization for illumination levels or placement of light sources was observed among the15 security checkpoints visited. According to Vine (1982), dim lighting encourages slowmovement and lethargic labor. Vine noted that simply replacing dark colors with bright shadesand increasing lighting levels to at least 100 foot-candles is reported to have improved processingrates the movement of people by 30 percent.

At two checkpoints, both located in an atrium area with glass canopies, severe glare fromsunlight was a problem at the X-ray screening positions. The image on the display monitorscould not be adequately seen. The degree of glare was severe enough to adversely impactscreener performance. At both of these checkpoints, screeners used pyramid-shaped shieldsattached to the face of the display monitors. Although successfully eliminating the glare, thesedevices had apertures that sharply restricted the operator's field of vision. Further design work isneeded to eliminate glare problems without affecting screener performance.

4.2 FLOOR GRADE.

One checkpoint was noted to have an estimated grade of 3 percent over the length of thecheckpoint area. All personnel, whether seated or prone, had to perform the job on an unevensurface. The effects on performance, or contribution to fatigue, need to be addressed in such awork environment. Optimally, this checkpoint could be reconfigured to eliminate the unevenfloor grade.

4.3 WORKSTATION ARCHITECTURE.

Almost 20 years ago, Astley and Fox (1975) indicated that a major ergonomics commitment wasneeded to redesign the physical dimensions of the industrial inspection workplace usinganthropometric and biomechanical data. Nonetheless, several workstations were constructedsuch that the screener's view of approaching passengers was obstructed. In some cases theobstruction was caused by placement of the screening units next to permanent structures (i.e.,walls, support pillars) of the facility, but in others, the obstruction was caused by the actualdesign of the checkpoint station.

This was particularly true of the so-called "high technology" checkpoints. High technologycheckpoints often had portable walls placed between the X-ray position and the entry side of thescreening unit. The view of approaching individuals was completely obstructed in most cases.This disadvantage requires that either another screener orally initiate activating the conveyor belt,or that the belt is run continuously during the shift. Typically the screening unit was keptcontinuously operational. Continuous operation demands that the screener maintain vigilance onthe display monitor throughout the shift. The literature is rich with data that demonstratesvigilance declines over a 30-minute interval. It is critical to ensure that constant vigilance is notrequired over 30-minute intervals.

32

Allowing the view of oncoming passengers permits a screener to activate the X-ray screeningdevice only when needed to screen carry-ons. This permits the screener numerous periods ofinactivity (i.e., rests) and decomposes the shift into a multitude of short sessions of operations.Vigilance is then only required during brief periods of operations. Additionally, the screener hasdirect expectations of when to maintain vigilance. Requiring screeners to maintain constantvigilance on the display over the entire shift could easily cause partial or full images to pass byunseen.

The recommended solutions to these design flaws are to: (a) replace opaque walls withtransparent Plexiglas structures; (b) install convex mirrors near the entry portals to permit view ofoncoming passengers; or 3) install weight-activated micro-switches on the conveyor belts toactivate the X-ray scanning device, with a visual or auditory cue to alert the screener to deviceactivation. The first two solutions are cost-effective and would reduce screener workload.

A similar problem was noted with regard to the design of bag check stations. At some hightechnology checkpoints portable partitions obstructed the view of bag check stations from the X-ray station. This design flaw prevents screeners from seeing the results of a requested bag check.An important element of any training program is the use of feedback to develop expertise andproficiency. This is vital in developing X-ray scanning proficiency in screener personnel.Feedback is necessary for screeners to develop skills in identifying objects from their X-rayimages. Repetition in relating X-ray images to the visual sight of objects is an important tool forskill development. When this feedback loop is removed an important training technique is lost.It is recommended that bag check stations are always in full view of the X-ray scanning positionto make available the opportunity for training.

4.4 COMMUNICATION AND DATA TRANSFER SYSTEMS.

High technology checkpoints characteristically linked X-ray screeners with CSS personnel usingheadsets. An additional feature of these checkpoints includes' that transmission of the X-rayimages from each checkpoint X-ray display monitor to a central information center staffed by theCSS. We observed that the CSS could be viewing as many as four X-ray display monitorssimultaneously on four individual displays. These monitors could also display information fromremote video cameras, and not necessarily be receiving images from the operational X-raymonitors. Screeners operating the X-ray scanning position were not in visual contact with theCSS and would not be aware of what, if any, information was being received by the CSS fromtheir individual monitor.

The consequences of these design features are multi-faceted. Issues regarding the diffusion ofresponsibility for screening carry-ons need to be addressed to determine if X-ray screeners areless vigilant in their performance if they believe a supervisor is also screening from a remotelocation. The impact of isolating the CSS from screeners also needs to be explored with regardsto the affects on training effectiveness, team cohesiveness, supervisory contact, motivation, andjob competence. We observed that the visual isolation created by the architecture of thesesystems was not conducive to building a team approach to security. X-ray positions wereisolated from viewing approaching passengers, bag check stations, and visual contact with

33

supervisors. In short, the X-ray screener was limited in their situational awareness of the entiresecurity environment.

4.5 FURTHER CONSIDERATIONS.

Several additional features of airline security screening checkpoints were found to warranthuman factors engineering intervention (see table 5). Among these were the need to address: (a)the seating provided for the X-ray scanning position; (b) the ambient noise levels aroundcheckpoints and the need for sound absorption materials; (c) the effect of night work andcircadian rhythms on screening performance; (d) the need for training of handling and liftingheavy loads; and (e) good management training to supervisors (e.g., TQM). These are only a fewof the issues that warrant attention. The investigator is referred to Grandjean's (1980) "Checklistfor the Analysis of the Workplace" and MIL-STD-1472C "Human Engineering Design Criteriafor Military Systems, Equipment and Facilities" for a comprehensive treatment.

TABLE 5. WORKSTATION EVALUATION AT SAN FRANCISCOINTERNATIONAL AIRPORT (SFO)

CONCOURSEA B C D E F

Floor grade Y Y Y Y N N(steep) (2 percent) (2 percent)

Noise level N N N N Na NGlare problem(s) N N NbN N Nb NLighting level Low N N N Lowc NVisibility/Obstructions N N N N N NDistance to exit N N N N Na Y

(30 yds)Space Y N N N N N2-Stage Mag N N N Y N Y"Read" position Stand Stand Stand Stand Stand SitClimate N N N N N N

(Y = problem, N = no problem)

a Moderate level of vibration under stations by joiner beam that runs width of checkpoint;

high frequency nature.b Late afternoon through rotunda.c Glare by type of lighting installed.

These issues were presented to highlight some of the problematic areas related to securitycheckpoint design. The significance of these factors is a concern for the impact they present onhuman performance. In a work environment where fatigue and vigilance decrements can rapidlydegrade performance and system efficiency, these issues present challenges that require attention.

34

It is also plausible that a poor work environment may contribute to decreased job longevityamong screeners. Most of the issues presented can be resolved with minimal investment. Weagree with other investigators (Drury and Fox, 1975) who stated, "... inspection has proved anexciting vehicle for ergonomics" (p. 299). Their observation is quite accurate based upon ourinitial findings.

35

5. OTHER SOURCES.

1. Adams, J.A., Humes, J.M. and Stenson, H.H. (1962). Monitoring of Complex VisualDisplays: III. Effects of Repeated Sessions on Human Vigilance. Human Factors, 4:149-157.

2. Astley, R.W., and Fox, J.G. (1975). Human Reliability in Quality Control. The Analysisof an Inspection in the Rubber Industry. C.G. Drury and J.G. Fox (Eds.). New York:Taylor and Francis, pp. 253-272.

3. Atkinson, A.P.C. (1973). Selection of the Necessary but Not Sufficient Skills for a Job.Human Factors, 15(2): 125-128.

4. Bakan, P., Belton, J.A., and Toth, J.C. (1963). Vigilance: A Symposium. Extraversion-Introversion and Decrement in an Auditory Vigilance Task. D.N. Buckner and J.J.McGrath (Eds.). New York: McGraw-Hill.

5. Berch, D.B., and Kanter, D.R. (1984). Sustained Attention in Human Performance J.Warm (Ed.). Chichester, UK: Wiley, pp. 144-178.

6. Bevan, W., and Turner, E.O. (1965). Vigilance Performance with a Qualitative Shift inVerbal Reinforcers. Journal of Experimental Psychology, 70: 83-86.

7. Boff, K.R., and Lincoln, J.E. (1988). Engineering Data Compendium: HumanPerception and Performance. AAMRL, Wright-Patterson AFB, OH.

8. Brennan, P.J. (1974). Job Satisfaction: Is There a Trend? Washington, DC: U.S.Government Printing Office.

9. Broadbent, D.E. (1958). Perception and Communication. New York: Pergamon.

10. Buckner, D.N., and McGrath, J.J. (1963). Vigilance: A Symposium. A Comparison ofPerformance on Single and Dual Sensory Mode Vigilance Tasks. D.N. Buckner and J.J.McGrath (Eds.). New York: McGraw-Hill.

11. Carmody, D.P. (1985). Eye Movements and Human Information Processing. FreeSearch, Restricted Search, and the Need for Context in Radiologic Image Perception. R.Groner, G.W. McConkie, and C. Menz (Eds.). Elsevier, North Holland, pp. 341-355.

12. Childs, J.M. (1976). Signal Complexity, Response Complexity, and Signal Specificationin Vigilance. Human Factors, 18(2): 149-160.

13. Coninsx, M. (1994). Seminar in Asia/Pacific Region Focuses on Role of Human Factorsin Aviation Security. Knowles, A. (Ed.). ICAO Journal, 49: 16-18.

36

14. Craig, A. (1984). Sustained Attention in Human Performance. Human Engineering: TheControl of Vigilance. J. Warm (Ed.). Chichester, UK: Wiley, pp. 247-291.

15. Craig, A., and Colquhoun, W.P. (1975). Human Reliability in Ouality Control.Vigilance: A Review. C.G. Drury and J.G. Fox (Eds.). New York: Taylor and Francis.

16. Craig, A., and Colquhoun, W.P. (1977). Vigilance: Theory, Operational Performance,and Physiological Correlates. Vigilance Effects in Complex Inspection. R.R. Mackie(Ed.). New York: Plenum, pp. 239-262.

17. Crumley, L.M., Pierce, L.G., Schwalm, R.C., Coke, J.S., and Brown, J.C. (1992).Predicting Target Detection Performance Using the Armed Services Vocational AptitudeBattery Subtests and Cognitive Factor Tests. (ARI Research Note 92-56). Alexandria,VA: U.S. Army Research Institute.

18. Davies, D.R. (1979). Search and the Human Observer. Theories of Vigilance andSustained Attention. M.A., J.N. Clare (Eds.). London: Taylor and Francis, pp. 14-19.

19. Davies, D.R., and Parasuraman, R. (1982). The Psychology of Vigilance. AcademicPress: London.

20. Davies, D.R., and Tune, G.S. (1970). Human Vigilance Performance. London: StaplesPress.

21. Drury, C.G., and Fox, J.G. (1975). Human Reliability in Quality Control. The Inspectorin the Future: His Efficiency and Well-Being. C.G. Drury and J.G. Fox (Eds.). NewYork: Taylor and Francis, pp. 299-301.

22. Drury, C.G., and Sheehan, J.J. (1969). Ergonomic and Economic Facts-an IndustrialInspection Task. International Journal of Production Research, 7: 333-341.

23. Edwards, D. (1990). Pilot Mental and Physical Performance. Ames, IA: Iowa State.

24. Eysenck, M.W. (1982). Attention and Arousal: Cognition and Performance. Berlin:Springer-Verlag.

25. Eysenck, M.W. (1983). Stress and Fatigue in Human Performance. Incentives. G.R.J.Hockey (Ed.). Chichester: Wiley.

26. Federal Aviation Administration. (1991, February). Semiannual Report to Congress onthe Effectiveness of the Civil Aviation Security Program. (Tech. Rep. DOT/FAA/ACS-89-1[3 1]). Washington, DC: NTIS.

37

27. Fox, J.G. (1975). Human Reliability in Ouality Control. Vigilance and Arousal: A Keyto Maintaining Inspectors' Performance. C.G. Drury and J.G. Fox (Eds.). New York:Taylor and Francis.

28. Frasher, R.G., and Pare, J.A.P. (1977). Diagnosis of Diseases of the Chest.Philadelphia: Saunders.

29. Funke, D.J. (1979, September). Security Management. Perceptual Training SharpensBaggage Inspection Skills, pp. 30-34.

30. General Accounting Office. (1987a, April). FAA Preboard Passenger Screening TestResults. (GAO/RCED-87-125FS): Washington, DC.

31. General Accounting Office. (1987b, July). Aviation Security: FAA Needs PreboardPassenger Screening Performance Standards. (GAO/RCED-87-182): Washington, DC.

32. Grandjean, E., and Koji, K. (1971). Methodology in Human Fatigue Assessment.Introductory Remarks. K. Hashimoto, K. Koji, and E. Grandjean (Eds.). London: Taylorand Francis.

33. Hertzberg, F. (1966). Work and the Nature of Man. Cleveland: World Publishing.

34. Howell, W.C., Johnston, W.A., and Goldstein, I.L. (1966). Organizational Behavior andHuman Performance. Complex Monitoring and Its Relation to the Classical Problem ofVigilance, 1: 129-150.

35. Howland, D., and Wiener, E.L. (1963). Vigilance: A Symposium. The System Monitor.D.N. Buckner and J.J. McGrath (Eds.). New York: McGraw-Hill, pp. 217-223.

36. Hughes, D. (1993, June 7). Aviation Week and Space Technology. FAA PreparesTesting Procedure for Explosive Detection Systems, pp. 122-125.

37. ITS (1993, April). The Personnel Performance Study. ITS Training and PersonnelDevelopment Department. Cleveland, OH: Author.

38. Jerison, H.J., and Wing, J.F. (1963). Vigilance: A Symposium. Human Vigilance andOperant Behavior. D.N. Buckner and J.J. McGrath (Eds.). New York: McGraw-Hill, pp.34-38.

39. Kaempf, G., Klinger, D., and Wolf, S. (1994). Development of Decision-CenteredInterventions for Airport Security Checkpoints. (Technical Report: DTRS-57-93-C-00129). Fairborn, OH: Klein Associates, Inc.

40. Keeping the Operator in Line. (1992, June). Airport Support, p. 41.

38

41. Kelly, M.L. (1955). A Study of Industrial Inspection by the Method of PairedComparisons. Psychological Monographs, 69: 1-16.

42. Kjellberg, A. (1977). Vigilance: Theory, Operational Performance, and PhysiologicalCorrelates. Sleep Deprivation, Arousal, and Performance. R.R. Mackie (Ed.). NewYork: Plenum.

43. Krupski, A., Raskin, D.C., and Bakan, P. (1971). Physiological and PersonalityCorrelates of Commission Error in an Auditory Vigilance Task. Psychophysiology, 8:304-311.

44. Levine, J.M. (1966). The Effects of Values and Costs on the Detection and Identificationof Signals in Auditory Vigilance. Human Factors, 8: 525-537.

45. Linestone, H. (1975). The Delphi Method: Techniques and Applications. Eight BasicPitfalls: A Checklist. H. Linstone, and M. Turoff (Ed.). New York: Addison WessleyPublishing.

46. Loeb, M., and Alluisi, E.A. (1984). Sustained Attention in Human Performance.Theories of Vigilance. J. Warm (Ed.). Chichester, UK: Wiley, pp. 179-205.

47. Mackie, R.R. (1987). Vigilance Research-Are We Ready for Countermeasures?Human Factors, 29(6): 707-723.

48. Mackworth, N.H. (1950). Researches on the Measurement on Human Performance.(MRC Report Series No. 268, Cambridge).

49. Mackworth, N.H. (1957). Some Factors Affecting Vigilance. Advanced Sciences, 53:389-393.

50. Mackworth, J.F. (1965). Effect of Amphetamine on the Detectability of Signals in aVigilance Task. Canadian Journal of Psychology, 19(2): 104-110.

51. McGrath, J.J., Harabedian, A., and Buckner, D.N. (1960). Human Factors Problems inAnti-Submarine Warfare. (Tech. Rep. No. 4). An Exploratory Study of the Correlates ofVigilance Performance. Los Angeles: Human Factors Research.

52. Meister, D. (1989). Conceptual Aspects of Human Factors. Baltimore, MD: JohnHopkins University Press.

53. Moder, J.J., and Oswalt, J.H. (1959). An Investigation of Some Factors Affecting theHand Picking of Small Objects. Journal of Industrial Engineering, 10: 213-218.

54. Murrell, K.F. (1965). Human Performance in Industry. New York,: Reinhold, Inc..

39

55. Nachreiner, F. (1977). Vigilance: Theory, Operational Performance, and PhysiologicalCorrelates. Experiments on the Validity of Vigilance Experiments. R.R. Mackie (Ed.).New York: Plenum, pp. 665-678.

56. Nadler, E.D., and Mengert, P. (1993). Safe Passage Data Analysis: Interim Report.(Tech. Rep. No. DOT-VNTSC-FA3T5-PM-93-3). Cambridge, MA: DOT-FAA.

57. Nelson, J.B., and Barany, J.W. (1969). A Dynamic Visual Recognition Test for PacedInspection Tasks. Transactions of Industrial Engineering, 1: 327-332.

58. Oster, C.V., Strong, J.S., and Zorn, C.K. (1992). Why Airplanes Crash. New York:Oxford.

59. Parasuraman, R. (1986). Handbook of Perception and Human Performance: Vol. 1I:Cognitive Processes and Performance. Vigilance, Monitoring, and Search. K.R. Boff, L.Kaufman, and J.P. Thomas (Eds.). New York: Wiley, pp. 1-39.

60. Pierce, L.G., Crumley, L.M., and Clifford, P.A. (1991). Detecting People Who CanDetect Targets. Proceedings of the Human Factors Society 35th Annual Meeting, 1005-1009.

61. Potter, A.N. (1975). Is Anti-Hijacking Success Only Temporary? Security World, 12: p.34.

62. Poulton, E.C. (1973). The Effects of Fatigue Upon Inspection Work. AppliedErgonomics, 4: 73-83.

63. President's Commission. (1990). Report to the President by the President's Commissionon Aviation Security and Terrorism. Executive Order 12686, August 4, 1989:Washington, DC. Sheehan, J.J., and Drury, C.G. (1971). The Analysis of IndustrialInspection. Applied Ergonomics, 2: 74-78.

64. Sipowicz, R.R., Ware, J., and Baker, R.A. (1962). The Effects of Reward andKnowledge of Results on the Performance of a Simple Vigilance Task. Journal ofExperimental Psychology, 64: 58-61.

65. Smith, R.L., Lucaccini, L.F., and Epstein, M.H. (1967). The Effects of MonetaryRewards zind Punishments on Vigilance Performance. Journal of Applied Psychology,51: 411-416.

66. Smith, R.L., and Lucaccini, L.F. (1969). Vigilance Research: It's Application toIndustrial Problems. Human Factors, 11: 149-156.

67. St. John, P. (1991). Air Piracy, Airport Security, and International Terrorism. NewYork: Quorum Books.

40

68. Study Tallies Hostile Acts against Civil Aviation. (1991, October). Air Line Pilot, p.48.

69. Sutton, D. (1969). A Textbook of Radiology. Baltimore, MD: Williams and Wilkins.

70. Thackray, R.I., Bailey, J.P., and Touchstone, R. (1977). The Effect of IncreasedMonitoring Load on Vigilance Performance Using a Simulated Radar Display. (Rep. No.FAA-AM-77-18). Federal Aviation Administration, Washington, DC: Office ofAviation Medicine.

71. Thackray, R.I., Jones, K.N., and Touchstone, R.M. (1974). Personality andPhysiological Correlates of Performance Decrement on a Monotonous Task RequiringSustained Attention. British Journal of Psychology, 65: 351-358.

72. Thaher, F.F. (1991). A Study of Security Safeguards at King Abdulaziz InternationalAirport in Jeddah, Saudi Arabia. Unpublished Graduate Research Project, Embry-RiddleAeronautical University, Daytona Beach, FL.

73. Tune, G.S. (1966). Errors of Commission as a Function of Age and Temperament in aType of Vigilance Task. Journal of Experimental Psychology, 18: 358-361.

74. U.S. Department of Defense. (1981). Human Engineering Design Criteria for MilitarySystems, Equipment and Facilities. (MIL-STD-1472C). Redstone Arsenal, AL: Author.

75. Vickers, D., Nettelbeck, T., and Willson, R.J. (1972). Perceptual Indices ofPerformance: The Measurement of "Inspection Time" and "Noise" in the Visual System.Perception, 1: 262-295.

76. Vine, D. (1982, November). Airport Services Management. Airport Security Tips (ForThose New to the Screening Game). pp. 24-25.

77. Vosburgh, A.R. (1993). Analysis of Some Aspects of Aviation Security TodayConsidered against Factors in the 1988 Bombing of Pan Am Flight 103. UnpublishedGraduate Research Project, Embry-Riddle Aeronautical University, Daytona Beach, FL.

78. Wallis, R. (1992, June). Jane's Airport Review. U.S. Proposes Criminal BackgroundChecks, pp. 21-26.

79. Warm, J.S., Dember, W.N., Lanzetta, T.M., Bowers, J.C., and Lysaght, R.J. (1985).Information Processing in Vigilance Performance: Complexity Revisited. Proceedingsof the Human Factors Society 29th Annual Meeting, 1: 19-22.

80. Wickens, C.D. (1992). Engineering Psychology and Human Performance, 2nd edition.New York: Harper Collins.

41

81. Wiener, E.L. (1969). Money and the Monitor. Perceptual and Motor Skills, 29: 627-634.

82. Wiener, E.L. (1975). Human Reliability in Quality Control. Individual and GroupDifferences in Inspection. C.G. Drury and J.G. Fox (Eds.). London: Taylor and Francis,pp. 101-122.

83. Wiener, E.L. (1987). Application of Vigilance Research: Rare, Medium, or Well Done?Human Factors, 29(6): 725-736.

84. Williams, L.G. (1966). Target Conspicuity and Visual Search. Human Factors, 8(2):80-91.

85. Wylie, C.D., Mackie, R.R., and Smith, M.J. (1985). Stress and Sonar OperatorPerformance: A Literature Review and Operator Survey. (Tech. Rep. 483-1). Goleta,CA: Essex Corp.

42